hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
robb83 has joined #linux-sunxi
robb83 has quit [Client Quit]
akaizen_ is now known as akaizen
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<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)
<Tsvetan> wow
* mnemoc seconds Tsvetan's wow
<oliv3r> mnemoc: you missed that last night?
<mnemoc> yeah
<oliv3r> who worked on this again, jemk, wingrime and nove, right?
<oliv3r> i mostly remember these 3 names
<mnemoc> i know i added jemk to linux-sunxi's cedarx team in github
<oliv3r> wingrime should be there at the very least ;)
<mnemoc> nove not there.... account?
<oliv3r> i'm not sure
<oliv3r> nove == paulo?
<Tsvetan> what about hardware encoding? is there wip for this too?
<oliv3r> far from :p
<oliv3r> encoding is much harder, since we dont' hae demo code
<mnemoc> there are some webcam a20-based devices for conferencing
<Tsvetan> oliv3r can you check why in the new Kernel 3.4 AW_IRQ_GPU_PMU and AW-IRQ_TOUCH_PANEL are set both to (AW_IRQ_GIC_START +29)
<Tsvetan> this seems is the reason touch panel doesn work with the new kernel
<Tsvetan> mnemoc we have streamed video too with A20 using just software
<Tsvetan> Dimitar made robot with web cam on it and it moves around and stream the video
<Tsvetan> but without hardware encoding
<oliv3r> Tsvetan: could be SMP IRQ offset
<oliv3r> Tsvetan: though it should be atleast +32 i think; i saw a patch regarding this pass the ML
<mnemoc> Tsvetan: can you ask aw for example libve code for encoding?
<oliv3r> Tsvetan: every core gets 32 'own' IRQ's I belive
<oliv3r> mnemoc: good point yes, jpeg, mpeg and h264 encoding examples would be perfect
<Tsvetan> mnemoc will do
<oliv3r> as those are required for register traces
<oliv3r> mnemoc: can we embed youtube on the wiki?
<mnemoc> i suppose so, you can embed html
<oliv3r> mnemoc: how do I embed html?
<oliv3r> the [html] tag doesn't work
<oliv3r> excellent
<mnemoc> choose one and I'll install it
<mnemoc> without opening the door to harmful html injections
<oliv3r> pbetter
<oliv3r> oh god, too much to choose :(
<oliv3r> this sounds reasonable
ykchavan has joined #linux-sunxi
<mnemoc> so be it
atiti has joined #linux-sunxi
<mnemoc> oliv3r: tells EmbedVideo (Version 1.0)
<oliv3r> ok let me refresh
<oliv3r> Status
<mnemoc> :)
<oliv3r> works
<oliv3r> anything you think we should change/reword/
<mnemoc> maybe reduce a bit the size of the video windown? :p
<mnemoc> i understand the excitement, but it's kind of a large video
<mnemoc> also "Reverse Engineering" doesn't only imply cedarx
<oliv3r> it's full hd video!
<oliv3r> fine, i'll make it smaller for you
<oliv3r> you mean the title of the page?
<oliv3r> erm, location
<mnemoc> i would put the video in a corner of the first section of the CedarX Reverse Engineering page
deasy has joined #linux-sunxi
<mnemoc> but.... you are the wiki master
<oliv3r> pff, i'm far from it
<oliv3r> i want it to be a seperated page
<mnemoc> ok
<oliv3r> mnemoc: reason for a seperate page it's better shareable with other resources :)
<mnemoc> you are the wiki master
<oliv3r> haha, far far from it
<oliv3r> :)
<oliv3r> lets see how this propagates over the intertubes
<mnemoc> and JIT a post in tizen's mailing list about allwinner support ranting about the VPU blob :p
<mnemoc> so propagation begins
<oliv3r> really?
<oliv3r> link?
<rm> mnemoc, I don't think he means VPU
<rm> just "video drivers" as in "videocard"
<rm> and "acceleration" of X11 and OpenGL ES
<mnemoc> rm: i only needed an excuse to propagate the video/status of cedarx's RE project :p
<rm> well what's wrong with propagating the status of Mali instead
<mnemoc> you can now reply pointing to ssvb's X11 drivers and lima
<rm> could have been closer to what is required by the user
<mnemoc> go ahead
<rm> I am not subscribed
<rm> actually
<rm> wait a second
<rm> he links to an A31 board
<rm> so no Mali there
<rm> does it still use the same CedarX?
<oliv3r> we believe so
jukivili has quit [Remote host closed the connection]
vicenteH has quit [Ping timeout: 260 seconds]
popolon has joined #linux-sunxi
jemk has joined #linux-sunxi
<oliv3r> Hello jemk
<oliv3r> hero!
<jemk> oliv3r: hello
<oliv3r> 8,250,514.2? 4,60EindhovenHelmondGemeente Helmond
<oliv3r> 16,7514.2? 4,60HelmondEindhovenGemeente Helmond
<oliv3r> 8,250,514.2? 4,60EindhovenHelmondGemeente Helmond
<oliv3r> 8,250,514.2? 4,60EindhovenHelmondGemeente Helmond
<oliv3r> 16,7514.2? 4,60HelmondEindhovenGemeente Helmond
<oliv3r> 16,7514.2? 4,60HelmondEindhovenGemeente Helmond
<oliv3r> erm
<oliv3r> whoops
<oliv3r> copy paste fail
<jemk> oliv3r: you wrote in the wiki "reverse enginered code", i didn't look at any code, only traces. so i don't think its that dark grey
<oliv3r> jemk: yeah but do we really want to risk anything?
<oliv3r> traces could be concidered reverse engineering
<oliv3r> and could be legally sensitive in some countries
<jemk> oliv3r: no, of course not, but how could one do it better? somehow i have to get information how things work
<oliv3r> jemk: nope you did it right, 'cleanroom' style
<oliv3r> you looked at traces, wrote 'documentation'
atiti has quit [Ping timeout: 260 seconds]
<oliv3r> now someone else who didn't touch any of that can take the documentation and write a driver/userspace
<oliv3r> and it's then cleanroom RE'd
<oliv3r> i don't think anything will ever happen
<oliv3r> t we don't want anybody using this code,s tart pushing it into gpl repostiories, and get a shitload of problems later
atiti has joined #linux-sunxi
<jemk> oliv3r: how does lima for example handle that?
sud0x3 has joined #linux-sunxi
<oliv3r> jemk: good question, i think libv does everything himself
<oliv3r> jemk: i suppose we'd have to get a lawyer to see what is right and what is wrong?
<Tsvetan> cedarx is written by allwinner, right?
<oliv3r> Tsvetan: we belive so yeah; it's really ugly piece of mess :)
<oliv3r> and the blob is a GPL violating one
<Tsvetan> I think the possibility they to chase someone who did RE on their code is low
<Tsvetan> but I can ask Wits/Allwinner about this
<mnemoc> they have been using the same VPU IP since ever
popolon has quit [Quit: Quitte]
mouchon has quit [Remote host closed the connection]
<Tsvetan> mnemoc I got email with promise for encoding demo code few minutes ago, let see when they will send it
<mnemoc> :D
<Tsvetan> as sometimes they forgot :)))
<oliv3r> Tsvetan: very cool :
<oliv3r> yeah they love to forget things
geecko has joined #linux-sunxi
<mnemoc> well... considering everything in that company is done by Eva, it's understandable she forgets things
<oliv3r> hahaha
popolon has joined #linux-sunxi
RaYmAn has joined #linux-sunxi
boycottg00gle has joined #linux-sunxi
<boycottg00gle> getting closer to my hdmi problem on a olimex a20: i2cdetect suggests hdmi is not connected but it is
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
vicenteH has joined #linux-sunxi
<oliv3r> Turl: good morning to you too!
<Turl> oliv3r: morning :)
<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 ?
<oliv3r> wingrime: nice!!
<boycottg00gle> hpd return 0 sounds like a hardware problem? (echo f1c1600c > /sys/class/misc/sunxi-dbgreg/rw/read => 0)
_enrico_ has joined #linux-sunxi
<mouchon> oliv3r yesterday i also tried different thing to change i2c bus interface but till now no luck :(
<oliv3r> boycottg00gle: i've onyl ever connected my cubie1 via hdmi, so i don't know how or what :(
<oliv3r> mouchon: have you tried to run it at 50Khz and checked with scope?
_BJFreeman has joined #linux-sunxi
<mouchon> oliver i have a logical analyser able to analyse i2c signal and clk is still at 200khz
_BJFreeman is now known as BJfreeman
<oliv3r> strange, it should be either 125 kHz or 40 kHz :)
<oliv3r> 400*
<oliv3r> but 400 / 8 is the 50 you want :)
jemk has quit [Ping timeout: 276 seconds]
<wingrime> oliv3r: but no ace driver in a20
<wingrime> sdk's kernel
<oliv3r> so you have blob in a20 sdk; but no cedarA kernel component?
<wingrime> yes
mouchon has quit [Ping timeout: 264 seconds]
<oliv3r> wingrime: tsvetsan is in contact with allwinner to get Encoding demo code :) so maybe he can get cedarA codec examples/blobs too
<oliv3r> wingrime: maybe cedarX kernel component works for both?
mouchon has joined #linux-sunxi
<wingrime> oliv3r: not sure, ace have other IO base
<oliv3r> wingrime: i mean, if there's ioctl that give full reg. access; then its possible
<mouchon> hargh i lost my connection
<wingrime> oliv3r: answer maybe simpler
<wingrime> oliv3r: a20 power enought to do it in soft
BluesBoy has joined #linux-sunxi
<oliv3r> wingrime: i think the audio engine is from the old sun3i days and maybe early sun4i days
<oliv3r> wingrime: but i bet the hardware still exists
<oliv3r> but we never found cedarA kernel module, so I would think it all does it via the register_write ioctl
<wingrime> oliv3r: we have ace driver in linux-sunxi
<wingrime> oliv3r: it even builds by default
<oliv3r> oh nice :)
<oliv3r> those linaro samplemedia does have audio stuffs
<wingrime> oliv3r: but I have no idea how make andoidish stuff builds for make traces
notmart has joined #linux-sunxi
<oliv3r> oh android, i dont' know that either
<oliv3r> libhybris around the cederA blob?
<wingrime> oliv3r: possible , but it have api around
BJfreeman has quit [Quit: had a good time]
notmart_ has quit [Ping timeout: 264 seconds]
<oliv3r> mouchon: are you back?
<mouchon> yes
<oliv3r> mouchon: you had the A20 that didn't boot from SD right?
<mouchon> resync of my dsl connection
<mouchon> yes indeed
<oliv3r> mouchon: assuming that your SD card is error free;
<oliv3r> dd if=u-boot-sunxi-with-spl.bin of=/dev/mmc0 seek=8 bs=1024
<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; cd u-boot-sunxi; A20-OLinuXino_MICRO CROSS_COMPILE=arm-linux-gnueabihf-
<mouchon> Tsvetan: may be on
stulluk has joined #linux-sunxi
notmart_ has joined #linux-sunxi
atiti has quit [Ping timeout: 260 seconds]
<stulluk> Hi, I have a strange issue, couldn't figure it out
<stulluk> I have an A20 board, and I mounted /dev/block/nanda to /mnt/nandA
<oliv3r> Tsvetan: rz2k your employee?
<stulluk> and I deleted script.bin file
<stulluk> it still can boot !
<n01> uhm is it safe to compile uboot with hf?
<stulluk> Is that normal?
<Tsvetan> oliv3r no
<oliv3r> stulluk: if you delete script.bin you are lost; it can't work anymore
<oliv3r> stulluk: you can try if you still have script.orig, which is a backup copy
<stulluk> oliv3r, funny point is, it still can boot
<oliv3r> n01: you can't, and i don't think u-boot supports floats, but there's no floats in u-boot anyway, so shouldl be safe yes
<stulluk> how this can happen?
<oliv3r> stulluk: without script.bin it's impossible to boot
<oliv3r> ohh CAN
<oliv3r> stulluk: ther'es a backup copy that's get parsed if script.bin is missing/corrupt
notmart has quit [Ping timeout: 260 seconds]
<n01> oliv3r: I was referring to CROSS_COMPILE= mouchon
<oliv3r> script.orig i belive
<oliv3r> n01: i compile u-boot with cross chain
<mouchon> Tsvetan: on you get 50G for free
<oliv3r> mouchon: nothing is even free
<oliv3r> ever*
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?
<oliv3r> yep
<oliv3r> script.bin and script.orig
<oliv3r> the chip can boot without it
<oliv3r> but a lot of perhipals may not work
notmart_ has quit [Ping timeout: 264 seconds]
<stulluk> Thats cool info
<stulluk> Thank you very much
<HeHoPMaJIeH> wingrime, you can download new SDK from here
<HeHoPMaJIeH> wingrime, tell me if you have any download speed issues
<stulluk> This SDK is newer than the old A20 torrent on olimex's website?
<HeHoPMaJIeH> yes it is
<stulluk> Cool
<stulluk> Thank you very much
<stulluk> just started download now
<stulluk> Is there a changelog ?
<oliv3r> lol
<HeHoPMaJIeH> i don't think so
<Tsvetan> :-)
<mouchon> just try with CROSS_COMPILE=arm-linux-gnueabi- but same issue. May be the chain i use on ubuntu 12.04 ?
<wingrime> oliv3r: may be better do it too wget to our server?
jemk has joined #linux-sunxi
<stulluk> Download speed 400KBps -- eta 4 hours 44 mins
<HeHoPMaJIeH> ;)
<oliv3r> mouchon: make sure you do a git pull ;)
<oliv3r> i used gcc 4.6.3
<mnemoc> is the whole channel downloading the file at the same time? :p
<mouchon> oliv3r: same version, git pull result "Already up-to-date"
<shineworld> :)
<Tsvetan> dimitar was downloading it yesterday from china with 4K/sec :)))
<wingrime> HeHoPMaJIeH: can you make costrated version, we actualy not need full android
ZetaNeta has joined #linux-sunxi
<mnemoc> Tsvetan: 4K and only one day?
<oliv3r> mouchon: what repo are you using?
<Tsvetan> lichee is 1G only
<stulluk> Speed reduced to 100Kbps and slowing down
<mnemoc> Tsvetan: :)
<Tsvetan> mnemoc - two days actually
<oliv3r> torrent would be much more usefull for this kind of stuff :)
<shineworld> 2h then I will share on another place for you
<stulluk> +1
<wingrime> 100 kb/s
<wingrime> jemk: ping
<shineworld> 700K
<oliv3r> mouchon: that's the right one
<mnemoc> 150K/s eta:12h .... i'm winning!
<oliv3r> well you have the same code, so you should get the same binary :(
<HeHoPMaJIeH> :D
<mouchon> oliv3r:branch sunxi
<oliv3r> mouchon: i don't know why you can't compile a working u-boot :(
<wingrime> oliv3r: also I have noticed in ace driver that ace realy have some modules
<oliv3r> :)
<mouchon> i will restart from scratch :-)
<oliv3r> PNG??
<HeHoPMaJIeH> keep in mind that host has only 10Mbit upload channel
<wingrime> ......
<oliv3r> HeHoPMaJIeH: torrent would have worked better ;)
<wingrime> HeHoPMaJIeH: you can make prvate torrent
<mnemoc> in 12h it will be in too
<mnemoc> and then should be able to join the torrent seeding
<HeHoPMaJIeH> if i make torrent the results will be pretty same, only i have this file :)
<oliv3r> HeHoPMaJIeH: not true :)
<wingrime> HeHoPMaJIeH: no realy, torrent have block randomizaton, so we all have max speed
<oliv3r> HeHoPMaJIeH: you will share piece A to user A, piece B to user B and piece C to user C
<oliv3r> but then, user A can share piece A to users B and C
<oliv3r> so it will speed up conciderably quickyl
<wingrime> oliv3r: about PNG, why not, sun3i are PMP
<oliv3r> wingrime: no, but png is good to have
<oliv3r> wingrime: but how does it fit into cedar AUDIO :)
<stulluk> HeHoPMaJIeH: Can you check the SDK, if they added Alsa-lib and Alsa-utils this time ?
<wingrime> oliv3r: see
<wingrime> oliv3r: ACE - have AE and CE
<wingrime> wtf
<HeHoPMaJIeH> stulluk, you mean if it is added to buildroot package ?
<stulluk> no, in Android external packages
<stulluk> In old SDK
<stulluk> they didn't added many important external packages
<stulluk> such as i2C tools and Alsa utils
<stulluk> I added I2C tools by myself easily
<stulluk> but Alsa-lib and alsa-utils are not so easy to add
<stulluk> So I wonder, if they added this time or not
<stulluk> In old SDK, they added Tinymix, but it was not compiled&copied to target RootFS though
<stulluk> So I wish they fixed it this time hopefully
<HeHoPMaJIeH> stulluk, you have no luck :)
<stulluk> oh my god..
<stulluk> May other chipset vendors are adding these... Why Aw not...
<stulluk> I am working on Imx6, TI, Amlogic...ect SDKs, almost all of them add such things as default....
<stulluk> HeHoPMaJIeH: Did they update busybox hopefully? or still 1.18.3 with many important missing scripts ? such as FTPGET FTPPUT ?
<HeHoPMaJIeH> stulluk, i am not examine it in details, android is not my OS :)
<stulluk> I understand, I also hate Android, I am also linux lover, but unfortunately for my project need to use android...
<wingrime> oliv3r: wtf TSCC is Techsmith Codec ??/
<stulluk> IF possible, I will try to build a chrooted environment with this A20 device, but need time..
<HeHoPMaJIeH> mnemoc, i have some strange irq issues with sun4i-ts in stage/sunxi-3.4
<HeHoPMaJIeH> it seems to be irq assigning problem but i can't figure it out
<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...
<HeHoPMaJIeH> we will understand soon :)
<oliv3r> mouchon: gcc 4.6.3
<oliv3r> wingrime: wth is techsmich codec?!
<wingrime> oliv3r: I have say wtf
<oliv3r> some screen capturing codec
<oliv3r> wingrime: quite commonly used appearatnly
<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
<mripard> oliv3r: => a bit more than 2 weeks.
<oliv3r> mripard: rc7 will be the last one?
<mripard> except something big comes up, yes
<stulluk> oliv3r: I checked almost everywhere in the old SDK... there is no boot.scr, no boot.env file or no boot.env partition...
<stulluk> I just can't figure out the structure of this build environment,
<stulluk> When I do this: ./ -p sun7i_android
<oliv3r> stulluk: i've never used any of the AW build enviroments :) i wouldn't touch it with a 10 foot pole :)
<stulluk> As I understand, it compile&link kernel & uboot
<stulluk> Are there anybody familiar with AW android SDK here?
<stulluk> oliv3r: Thank you for trying help at least...
<oliv3r> stulluk: thank Allwinner ;)
<oliv3r> mripard: if greg doesn't respond to your query whether to use stable/testing ABI; should I better move it to testing for the final version
<mripard> yep, I think so
<oliv3r> rgr
<oliv3r> done
techn_ has joined #linux-sunxi
vicenteH` has quit [Ping timeout: 276 seconds]
nove has joined #linux-sunxi
<nove> 06:50 <oliv3r> nove == paulo?
<nove> oliver, you failed to guess again, i not /used/am using/will use/ multiple identities.
<oliv3r> nove: haha, i often fail to guess ;)
<oliv3r> nove: but I also get names wrong often!
<nove> I am only someone that saw the potencial of this socs to be "fully" open source, and wanted to help in the missing points.
<nove> And that was the why for tracer.
<wingrime> oliv3r: nove , your python visualizer eat my PC ram and not display simple 64 Mb trace
<oliv3r> nove: you helped, and that's that.
<oliv3r> nove: you are a hero :)
techn__ has quit [Ping timeout: 260 seconds]
<oliv3r> wingrime: who else worked on cedarX RE effort?
<oliv3r> i faintly remember a 4th name
<nove> wingrime, as expected, only as test with < 1M traces
<wingrime> oliv3r: nove , me , jemk , paullo612 (VP6)
<oliv3r> i'll add his name to the wiki
hipboi has joined #linux-sunxi
<nove> paullo612, the decompiled code?, is vp8
<oliv3r> hipboi: have you hear dthe latest news?
<nove> oliv3r, i only did the tracer
<wingrime> nove: yes
<wingrime> nove: and also added lots info about bp8
<hipboi> oliv3r, thank you
<nove> wingrime, jemk, good job
<hipboi> oliv3r, this is really good news
<andoma> \o/
<hipboi> oliv3r, who should we thank to
<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
<stulluk> I see, I also hate that PMIC
<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
<nove> wingrime, where is you python parser?
<wingrime> nove: wait a second
wingrime_ has joined #linux-sunxi
<mnemoc> oliv3r: they switched??!
<nove> wingrime, i am planing to add some patch to the libcedarv, to output to valgrind some info, like frames
<wingrime> nove: you can divide frames using "ioctrl 102 "
<wingrime> nove: thats mean 'Wait VE'
<HeHoPMaJIeH> mnemoc,
hansg has joined #linux-sunxi
<mnemoc> HeHoPMaJIeH: ow
<nove> wingrime, is that certain? doesn't the blob does
<wingrime> nove: blob ask kernel to IRQ wait
<nove> wingrime, anything more before display frame
<wingrime> nove: IRQ can be handled only by kernel
<oliv3r> mnemoc: appearantly HeHoPMaJIeH says so :)
<wingrime> nove: also decoding process are lauching with _TRIG registers
<wingrime> nove: but not only
<mnemoc> `/aw.tar.bz2' at 1985643856 (31%) 180.1K/s eta:6h47m
wingrime_ has quit [Ping timeout: 268 seconds]
<mnemoc> to
<oliv3r> mnemoc: 3.4.39+ maybe old sunxi kernel :p
<oliv3r> i doubt it
<mnemoc> i doubt it too
<oliv3r> looks like old crappy kernel
<mnemoc> i see no signs of our work there
<oliv3r> oh, maybe it has touchscreen driver for my tablet
<oliv3r> so same old bugs, AGAIN
<oliv3r> and with 3.4 name; it'll be really confusing to users
<oliv3r> mnemoc:have you extracted it yet?
<mnemoc> oliv3r: eta:6h13m
<oliv3r> HeHoPMaJIeH: can you check drivers/input/touchscreen and do a ls -laF somewhere
<mnemoc> find | sprunge
pitillo has quit [Ping timeout: 264 seconds]
<oliv3r> sprunge yes, or copy/paste :p
stulluk has quit [Ping timeout: 246 seconds]
ZetaNeta has joined #linux-sunxi
<oliv3r> HeHoPMaJIeH: ls -laF drivers/input/touchscreen | curl -F 'sprunge=<-'
<HeHoPMaJIeH> just a second
pitillo has joined #linux-sunxi
<HeHoPMaJIeH> oliv3r,
<oliv3r> HeHoPMaJIeH: thankyou
<HeHoPMaJIeH> mnemoc, result is too big to upload to sprunge
<oliv3r> that suggests still a lot of violating driversin the sdk
<oliv3r> well my TS still not in there :)
<shineworld> download finished
<mnemoc> HeHoPMaJIeH: ls -R | gzip > aw.txt.gz and to the same server?
<shineworld> md5 of file
<shineworld> ?
<mnemoc> bz2 will know if the file is corrupt
<HeHoPMaJIeH> find | gzip > /var/www/fl.gz
<shineworld> $ md5 aw.tar.bz2
<shineworld> MD5 (aw.tar.bz2) = 01330d458196ae203c7f7a31b90eaf64
<HeHoPMaJIeH> mnemoc, second try
<HeHoPMaJIeH> :)
<mnemoc> thanks :)
<HeHoPMaJIeH> np
atiti has quit [Ping timeout: 260 seconds]
<HeHoPMaJIeH> shineworld, 01330d458196ae203c7f7a31b90eaf64 /var/www/aw.tar.bz2
<shineworld> thanks so is fine ;)
<shineworld> time to put hands inner |
shineworld has quit [Remote host closed the connection]
ZetaNeta has quit [Ping timeout: 256 seconds]
kaspter has joined #linux-sunxi
<oliv3r> sometimes, comments are funny
nove has quit [Quit: nove]
hansg has quit [Quit: Leaving]
vicenteH has joined #linux-sunxi
igor- has joined #linux-sunxi
<wingrime> 5h 48m
ganbold_ has joined #linux-sunxi
<_enrico_> HeHoPMaJIeH: can you upload this dir: drivers/media/video/sun7i_tvd ?
<n01> _enrico_: fellow italian?
<_enrico_> n01: yes :D
<n01> wow ... small italian sunxi community here
geecko has quit [Quit: Quitte]
igor- has quit [Read error: Connection reset by peer]
igor- has joined #linux-sunxi
hipboi has quit [Read error: Connection reset by peer]
hipboi has joined #linux-sunxi
igor- has quit [Ping timeout: 245 seconds]
Soru has joined #linux-sunxi
notmart_ has joined #linux-sunxi
rellla has joined #linux-sunxi
notmart has quit [Ping timeout: 276 seconds]
kaspter has quit [Ping timeout: 264 seconds]
igor- has joined #linux-sunxi
hipboi_ has joined #linux-sunxi
hipboi_ has quit [Client Quit]
hipboi has quit [Ping timeout: 256 seconds]
rellla has quit [Quit: Nettalk6 -]
n01 has quit [Ping timeout: 264 seconds]
<wingrime> oliv3r:
<wingrime> oliv3r: spamers use Markcov Chain
<wingrime> Turl: how current spam protection system works?
notmart has joined #linux-sunxi
notmart_ has quit [Ping timeout: 240 seconds]
vinifr has joined #linux-sunxi
sud0x3 has joined #linux-sunxi
igor- has quit [Ping timeout: 256 seconds]
notmart has quit [Read error: Operation timed out]
notmart has joined #linux-sunxi
<lkcl_> has anyone managed to compile allwinner's release a20 u-boot code and get it to work with sd/mmc?
rings_IIV has joined #linux-sunxi
gnufan has quit [Ping timeout: 264 seconds]
sud0x3 has quit [Ping timeout: 245 seconds]
<techn_> ssvb: should neon handle 4 byte aligned addresses?
<techn_> .. without crashing
notmart_ has joined #linux-sunxi
notmart has quit [Ping timeout: 248 seconds]
Soru has quit [Read error: No route to host]
Soru has joined #linux-sunxi
sud0x3 has joined #linux-sunxi
igor- has joined #linux-sunxi
<ssvb> techn_: it depends, there are aligned load/store instructions and there are also unaligned
<ssvb> techn_: if one tries to use aligned instruction with an unaligned address, it will fail with SIGBUS
BluesBoy- has joined #linux-sunxi
<hramrach> wingrime: you have h264 working on Cedar?
<wingrime> hramrach: I make this video
<wingrime> jemk: ping
BluesBoy has quit [Ping timeout: 240 seconds]
BluesBoy- is now known as BluesBoy
<jemk> wingrime: yes?
<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_> dd if=spl/sunxi-spl.bin of=/dev/sdX bs=1024 seek=8
<techn_> dd if=u-boot.img of=/dev/sdX bs=1024 seek=40
<techn_> but still It wont notice bootable sdcard
_BJFreeman has joined #linux-sunxi
<Turl> techn_: hno moved stuff around recently, check the ML
_BJFreeman is now known as BJfreeman
notmart_ has joined #linux-sunxi
notmart has quit [Ping timeout: 260 seconds]
<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
<oliv3r> techn_: i don't recall us having it
<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: AW_VIR_TP_BASE is touchscreen?
<wingrime> oliv3r: my touchsceen
<oliv3r> 0x01c25000 0x01c253ff 1KiB TP Config registers Resitive Touch Panel controller
<oliv3r> yep
<oliv3r> 1c25000 is touchscreen controller
<oliv3r> so a10 has the same sensor
<oliv3r> that driver/file should work on a110
<oliv3r> i tried to read the registers via u-boot; but never got it to work
<oliv3r> maybe a clock issue?
<oliv3r> with a seperated driver, all clcoks shoudl be properly enabled
* Turl feels a disturbance in the force
<oliv3r> Turl: lol hi
<oliv3r> Turl: how does dt handle if you have 1 register block (touchpanel) with 5 registers, and register 3 is for the temperature sensor
<oliv3r> and you want a seperated driver for that
<oliv3r> 2 register blocks that skips that 1 register?
<wingrime> oliv3r: gemac get driver
<wingrime> *gmac
<oliv3r> nice one, but we had it in the a31 dump
<oliv3r> this will make it much easier though
<wingrime> oliv3r: gmac request new phy only?
<oliv3r> yeah
<oliv3r> there's 2 mac drivers in a20
<oliv3r> emac and gmac
<wingrime> oliv3r: and some wirings
<oliv3r> but they share the same pinout
<oliv3r> that's why cb1 is compatible to cb2
<oliv3r> even more interesting that the gmac has rgmii and gmii modes
<Turl> oliv3r: is the temp one needed for touch?
<oliv3r> and gmii is fully backwards compatible with mii
<oliv3r> so cb2 uses emac driver now (which is known to be buggy) but could potentially use the gmac driver, with the same mii
_enrico_ has quit [Ping timeout: 245 seconds]
<oliv3r> Turl: it might be used to calculate offsets and such, but no, in a20 it's a seperated driver
<oliv3r> Turl: a20 sdk has the driver split out; but, we know that Allwinner just bangs each register wherever it wants
<oliv3r> Turl: touchscreen driver currently works entirly without the sensor
<hno> techn_, that works for me. I assume you adjusted /dev/sdX accordingly.
<techn_> hno: strange.. It just wont detect bootable sdcard :(
<Turl> oliv3r: you can use the same register
BJfreeman has quit [Ping timeout: 268 seconds]
<Turl> I think you need to use of_iomap though, check with mripard
_enrico_ has joined #linux-sunxi
<Turl> and add locking if needed
<oliv3r> Turl: what about skipping that 1 register in the map; is that possible?
<oliv3r> mripard: ^ what's the cleanest way to use 1 register with driver a) and the surrounding 5 registers with driver B
<Turl> oliv3r: you could have 2 maps too, but imo not worth it
<wingrime> oliv3r: oh
<wingrime> oliv3r: nice
<wingrime> oliv3r: just see
<oliv3r> wingrime: ?
<wingrime> oliv3r: comparea this
<oliv3r> wingrime: commented code
<oliv3r> the same blob
<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
<mouchon> which os ?
<oliv3r> gentoo
<oliv3r> but that can't be the issue
<oliv3r> it cannot be so subtle
techn__ has joined #linux-sunxi
<oliv3r> what gcc versiona re you using?
<oliv3r> hno: ^ !!
<mouchon> Ubuntu/Linaro 4.6.3-1ubuntu5
<wingrime> Olaff: clk_set_rate(ve_moduleclk, pll4clk_rate/6); <- default clock
<wingrime> oliv3r:
<oliv3r> ok
<oliv3r> yeah
<oliv3r> let me compare
<rah> oliv3r: it seems you lie
<oliv3r> so 160 kHz
<oliv3r> rah: hahah, well not really; it's a really small wiki section :)
<wingrime> oliv3r: pll4/6 on sun4i
<oliv3r> and closed source, so we don't know much
<oliv3r> wingrime: that means 160 kHz though (according to the comment)
<oliv3r> wingrime: you think that's at ypo?
<wingrime> oliv3r: pll4/2 on sun5i
<mouchon> oliv3r configure used by ubuntu: ../src/configure -v --with-pkgversion='Ubuntu/Linaro 4.6.3-1ubuntu5' --with-bugurl=file:///usr/share/doc/gcc-4.6/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-4.6 --enable-shared --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/arm-linux-gnueabihf/include/c++/4.6.3 --libdir=/usr
<mouchon> /lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-gnu-unique-object --enable-plugin --enable-objc-gc --enable-multilib --disable-sjlj-exceptions --with-arch=armv7-a --with-float=hard --with-fpu=vfpv3-d16 --with-mode=thumb --disable-werror --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=arm-linux-gnueabihf --program-prefix=arm-linux-gnueabihf- --includedir=/usr/a
<oliv3r> so on sun4i it can run at 160, 240 or 320 kHz
<mouchon> rm-linux-gnueabihf/include --with-headers=/usr/arm-linux-gnueabihf/include --with-libs=/usr/arm-linux-gnueabihf/lib
<oliv3r> mouchon: i'm on 4.6.3 too
techn_ has quit [Ping timeout: 256 seconds]
<wingrime> oliv3r: but sun5i set pll4 too
<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 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
BluesBoy has quit [Quit: HydraIRC -> <- Now with extra fish!]
<hno> mouchon, how large is your spl/u-boot-spl.bin?
<wingrime> jemk: consider that a31 can run 4k video....
<mouchon> 24924
<wingrime> oliv3r: ohh
<wingrime> oliv3r: sun5i = 240 / 2 Mhz
<hno> mouchon, and how large is spl/sunxi-spl.bin?
<mouchon> 25088
<wingrime> oliv3r: sun4i = 720 / 6
<oliv3r> techn__: what hardware you running on?
<oliv3r> hno: i compiled olimex a20 u-boot for mouch that DID work
ZetaNeta has quit [Read error: Operation timed out]
<techn__> oliv3r: now I'm trying cubie2
<oliv3r> jemk: wingrime, eventually, dynamic clocking with VPU load monitoring
<wingrime> oliv3r: sun7i have default speed 300/1
<oliv3r> jemk: wingrime if VPU is failing to decode intime, increase clock (until known safe limit)
<wingrime> oliv3r: we can't know that
<jemk> wingrime: i cant see any difference
<wingrime> oliv3r: we can simply set clocks depend on scale
<hno> techn__ & mouchon, both of you get quite large SPL binaries. I'll triple check that large SPLs really work when I get home.
<wingrime> jemk: how big your video?
<hno> oliv3r, your compiler is making smaller SPL, just as mine.
<jemk> oliv3r: blob sets clock different depending on format, mpeg was lower than h264
<jemk> wingrime: 1920p sintel
<mouchon> ok
<oliv3r> hno: i still using the same compiler as always ..
<oliv3r> hno: the strange thing is, mouchin is using also linaro 4.6.3
<wingrime> jemk: witch profile?
<hno> oliv3r, how large is your spl/sunxi-spl.bin?
<jemk> wingrime: how to check?
<wingrime> ffprobe
<techn__> here is 4.7.2 ubuntu/linaro
<techn__> and cubieboard2 target
<mouchon> ubuntu too and olimex A20
<wingrime> jemk: currently a13 driver ignore freq request
<wingrime> jemk: marketing shit
<oliv3r> hno: u-boot-sunxi-with-spl.bin26095630-Aug-2013 09:52
<oliv3r> other files work, but are older
<oliv3r> or should work ;)
<wingrime> oliv3r: blob currently set freq to 240
<mouchon> hno: i also compiled on the board itself using the olimex r2 image
i-rinat has joined #linux-sunxi
<wingrime> oliv3r: consider that a13 ignore this call, looks likes a10's cedar run twice faster than a13's
<wingrime> oliv3r: thats why they write 2160 in a10 page
<hno> oliv3r, yours is < 24K.
<jemk> wingrime: 160clk -> ~30ms per frame 240clk -> ~22ms/frame 320clk-> ~17ms/frame
<i-rinat> jemk: Hi, just found your vdpau driver for cedarx. Here's another vdpau driver: 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 :)
<wingrime> techn__: witch size now?
<techn__> wingrime: 24108bytes
<wingrime> techn__: uboot.img ?
<techn__> wingrime: 171484
<wingrime> techn__: thast even can be fir in SRAM
<wingrime> *fit
lkcl_ has quit [Ping timeout: 245 seconds]
<jemk> wingrime: oups, that looks strange, but where is your mplayer window? did you do fullscreen?
<wingrime> no
<wingrime> jemk: actualy it even stranger
<wingrime> jemk: all windows shifted down
<wingrime> even mouse can reach to hi-part
<jemk> wingrime: aaahhh, thats disp bug/feature, it cant scale down that much, i think memory bandwidth issues
<jemk> wingrime: thats why blob prescales big videos with sdrot
<wingrime> jemk: but if I do fullscreen all normal
<wingrime> jemk: so with bandwidth all normal
<jemk> wingrime: i mean, to scale down you have to read much more in the short time an output line lasts
paulk-aldrin has quit [Quit: Ex-Chat]
<wingrime> jemk: I think using down scale register give us some speed-up
<wingrime> jemk: but I think it fixable without it
<jemk> wingrime: would be better to fix without sdrot, cause that would make decoding depend on window size, which vdpau doesn't allow
<wingrime> jemk: you have display object
<jemk> wingrime: but it looks like some issue with disp, i also have this when using small windows and noticed it with jpeg-poc too
<wingrime> jemk: if video much more than you screen , you can scale down
<wingrime> jemk: but indeed, in mem bandwidth term if can get some speed-up, but you have other buffer than output use same dram
<wingrime> jemk: and same size
ZetaNeta has quit [Read error: Connection reset by peer]
<techn__> hno: smaller spl is working
<oliv3r> that sucks kinda techn__ :)
<hno> techn__, ok. I'll reverify the size limits again.
wingrime_ has joined #linux-sunxi
wingrime has quit [Ping timeout: 245 seconds]
jemk has quit [Remote host closed the connection]
<n01> oliv3r: what dts are you using for cubie2
<oliv3r> its' from mripard tree with sid added :)
<n01> wow ... where this is from
<n01> ok
<oliv3r> actually
<oliv3r> in non-bin
<oliv3r> but that tree is very volatile, though I think of submitting it for v7 in its current form
<n01> ok it's not in sunxi-next in mripard's repo
<oliv3r> its based from his a20-clocks
<oliv3r> he said that was most up to date
<n01> I see
<oliv3r> Turl: what is the status of your mainline merge on linux-sunxi
<oliv3r> but mripard has lots of branchse
<oliv3r> i _think_ turls intention was to have a somewhat 'main' branch
techn_ has joined #linux-sunxi
<n01> hahah who posted on
techn__ has quit [Ping timeout: 245 seconds]
<oliv3r> probably propagated from phoronix
tinti has quit [Read error: Connection reset by peer]
wingrime_ has quit [Ping timeout: 256 seconds]
<oliv3r> Tsvetan: the opensource bootloader isn't really entirly contributed bya llwinner, hno did a lot of work with tom i belive
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 260 seconds]
<libv> i have es2gears against a mesa driver :)
<libv> will post video and more noise on monday :)
<libv> then i will do textures so i can run the other limare egl demoes, and then code-cleanup and pushing that out :)
n01 has quit [Ping timeout: 245 seconds]
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<Turl> libv: great! :)
<Turl> oliv3r: outdated these days, need to pick up all the A20/A31 patches mripard sent
_enrico_ has quit [Quit: Bye]
ykchavan has quit [Quit: Leaving]
torindel has quit [Ping timeout: 246 seconds]
sud0x3 has quit [Quit: Leaving]
torindel has joined #linux-sunxi
<pitillo> oliv3r: finally the problem was the card (smelled). Thank you :)