hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
<steev> arokux2: can i boot from sata?
<arokux2> steev, no idea
<steev> me either, that's why i asked
<arokux2> steev, you can boot from usb, what else do you want? :p
<arokux2> for example, usb external hard drive ;)
<steev> arokux2: sata > usb2
<arokux2> steev, how much MB/s?
<steev> good question
<arokux2> steev, USB - 27MB/s
<steev> uno momento
<steev> utilite ~ # hdparm -tT /dev/sda
<steev> Timing cached reads: 914 MB in 2.00 seconds = 457.00 MB/sec
<steev> Timing buffered disk reads: 348 MB in 3.01 seconds = 115.50 MB/sec
<steev> of course, that's the utilite, so it's kinda cheating, let me head over to the cb2 and plug in the sata drive
<arokux2> steev, ow.. you've got utilite, which model?
<steev> quad with 32gb msata ssd
<steev> 2gb ram
<steev> it's nice, very nice
<steev> very easily a desktop replacement
<steev> on the cb2, my intel 313 ssd shows 69MB/s
<steev> for buffered
<arokux2> steev, ok.. usb gets beaten
<arokux2> steev, does U-Boot support boot from SATA?
<arokux2> in general
<steev> it supports it on the utilite
<steev> but u-boot is stored in spi nor
<arokux2> steev, what kernel do you run on utilite?
<steev> it's a 3.0.35 base
<arokux2> steev, are you going to work on mainlining? :)
<steev> arokux2: i've debated it
<steev> as it stands, i've got 4 machines to get a linux distro running on, some gentoo work to catch up on, a new kernel to introduce into portage (which includes backporting support for the CuBox into it), plus my job
<arokux2> steev, what are you doing on your job?
<steev> currently, consolidating a bunch of machines that run different linux distros to all run on one, while also packaging up software that the distro doesn't run on
<steev> s/run on/package/
<arokux2> I see
<steev> also dodging companies that try to recruit me to come work for them :P
<arokux2> steev, what distro do you use at your work?
<steev> arokux2: personally i use gentoo
<arokux2> steev, that one I've got.
<steev> as for the consolidation, it's to get them all running on a customized debian base
<arokux2> steev, for me arch/debian are the best distros. gentoo is more hardcore than arch, but arch is flexible enough for me.
<steev> i'm not going to get into a distro war, everyone should use what they are most comfortable with
<arokux2> agree
<arokux2> I've struggled weeks before I added USB to mainline.
<arokux2> I've spent an evening to add it to U-Boot afterwards
<arokux2> next task - AXP209, PMIC.
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
arokux2 has quit [Remote host closed the connection]
<Turl> arokux1: once you learn the trick it's easy :p
<wens> arokux1: nice!
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<brain__> I think its time to give up on ever getting a working touch panel using this tablet with Linux, its great for what It is, and it was free with a loose CPU lol, but I need a more common A* based tablet that dont used this crap custom chip!
<brain__> I've tried everything short of sniffing the data from the touch panel lines using my bus pirate....which I'm leaving as last final resort before it makes contact with a wall hehe
<Turl> brain__: have you tried disassembling a working driver?
<brain__> I don't know which tools to use under linux. I have only peeked through using a hex editor to see if those custom strings which are not in standard code were there and see if it has integrated firmware which it does not.
popolon has quit [Quit: Quitte]
<brain__> custom strings were there tho, and its GPL V2 i just noticed, not sure if they were required to submit the changes they made
<brain__> any good disassemblers that work well under linux you'd recommend?
<Turl> brain__: yeah, they'd need to release their code if it's marked as GPLv2
<Turl> brain__: the practical odds of getting them are not usually that good though
<Turl> brain__: mailing the chip maker can't hurt I guess
<Turl> brain__: as for disassemblers, IDA is good
<Turl> brain__: you can use objdump if you like reading raw assembler :)
<brain__> IDA under linux? were have i been..
<brain__> im trying not to give up, I got it free broken, reflowed the CPU and installed a port on the side soldered into the internal uart..i just would like linux for my other projecst like arduino, SAM7, Beagleboard terminal, etc. wanna get as much use outta this thing..
<brain__> I'll check out IDA thanks
ganbold_ has joined #linux-sunxi
ganbold_ has quit [Remote host closed the connection]
ganbold_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
tm512` has joined #linux-sunxi
tm512 has quit [Ping timeout: 248 seconds]
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
pepee has joined #linux-sunxi
tm512` is now known as tm512
CoSM0 has quit [Read error: Connection reset by peer]
CoSM0 has joined #linux-sunxi
boss has joined #linux-sunxi
boss has left #linux-sunxi [#linux-sunxi]
\\Mr_C\\ has quit [Excess Flood]
\\Mr_C\\ has joined #linux-sunxi
hipboi has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
ykchavan has joined #linux-sunxi
hipboi has quit [Ping timeout: 252 seconds]
<Tsvetan> oliv3r FETs do not need current to operate but voltage
ykchavan has quit [Read error: No route to host]
uwe__ has joined #linux-sunxi
uwe_ has quit [Ping timeout: 264 seconds]
boss has joined #linux-sunxi
drachensun has quit [Ping timeout: 240 seconds]
drachensun_ has joined #linux-sunxi
hipboi has joined #linux-sunxi
\\Mr_C\\ has quit []
pepee_ has joined #linux-sunxi
ykchavan has joined #linux-sunxi
pepee has quit [Ping timeout: 252 seconds]
boss has quit [Ping timeout: 264 seconds]
eagles0513875__ has joined #linux-sunxi
<eagles0513875__> hey guys
<eagles0513875__> has support been added for the cubieboard 2 in the latest git version ?
<steev> of what
pepee_ has quit [Ping timeout: 268 seconds]
<eagles0513875__> build root steev last i tried it the cubieboard was not supported yet
<steev> i don't know what git that is, you could always look?
<steev> if you mean the sunxi-bsp, no
rellla has joined #linux-sunxi
<eagles0513875__> steev: found my answer
<eagles0513875__> seems like it was recently added
eagles0513875__ has quit [Ping timeout: 250 seconds]
FDCX_ has joined #linux-sunxi
atiti has joined #linux-sunxi
forcev has joined #linux-sunxi
FunkyPenguin has quit [Ping timeout: 276 seconds]
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
boss has joined #linux-sunxi
arokux2 has joined #linux-sunxi
n01 has joined #linux-sunxi
<arokux2> Tsvetan, didn't know if you backread. the cheapest Eagle license costs 160 EUR
bamvor_ has quit [Ping timeout: 248 seconds]
bamvor_ has joined #linux-sunxi
jeremb has quit [Ping timeout: 240 seconds]
jeremb_ has joined #linux-sunxi
massi has joined #linux-sunxi
boss has quit [Ping timeout: 264 seconds]
mcbrick has joined #linux-sunxi
bamvor_ has quit [Ping timeout: 240 seconds]
bamvor_ has joined #linux-sunxi
Williams323 has joined #linux-sunxi
rellla has quit [Quit: Nettalk6 -]
<Williams323> hi
Williams323 has left #linux-sunxi ["Ex-Chat"]
<Turl> morning
\\Mr_C\\ has joined #linux-sunxi
<oliv3r> Turl: mornin'
<oliv3r> arokux2: what was your big secret
<Turl> oliv3r: USB support in uboot :D
<oliv3r> that's done allready?
<oliv3r> that was fast :)
<oliv3r> Tsvetan: well voltage will be applied from the GPIO in any case :) but the lowest amperage we can supply is 10 mA so if tha'ts enough, that's good
<oliv3r> Tsvetan: but we'd need to know which revisions use that FET for sata_pwr_en for the devicetree :)
<Turl> btw I just pushed a fresh sunxi-devel, I did it early so I do not get behind later on :) just ping me when you send patches to mainline MLs and I'll add them to the tree on the spot
<hipboi> Turl, USB support in u-boot?
<hipboi> good news
<Turl> hipboi: yep :)
<Turl> hipboi: too bad we still need a sd to get uboot loaded
<hipboi> Turl, i think hno's fel boot can directly load from brom
<Turl> hipboi: yes it can, but it's not convenient for a "deployed"/"production" device
<Turl> still, it's cool that we can have rootfs and kernel on a usb drive now :)
Super-noob has quit [Ping timeout: 260 seconds]
Super-noob has joined #linux-sunxi
<arokux2> hipboi, yep, booting the kernel from USB Stick works. will push patches soon
Black_Horseman has quit [Ping timeout: 268 seconds]
<arokux2> oliv3r, see this and next message:;
<arokux2> Turl, will you plz CC linux-sunxi when sending your patches to mainline?
<Turl> arokux2: I always forget to do so :(
<Turl> arokux2: I also forgot to cc ojn
<arokux2> Turl, :( where have you sent your patches?
<Turl> arokux2: lakml
<Turl> arokux2: you can find them on the sunxi-devel I just pushed if you just want to have a look
<arokux2> Turl, I meant patches with PLL6
<Turl> arokux2: I haven't sent them yet
<arokux2> Turl, ok, so plz do not forget to CC us this time
<Turl> they're next in line on the pile :)
<Turl> arokux2: you should subscribe to lakml, you can learn a lot of stuff just by reading other people's patches
<arokux2> Turl, I am subscribed, but the filter I set automatically archives all e-mails from lakml :)
<arokux2> Turl, no time to read so much.... :( I just dig on stuff I'm working with. maybe this is not a good approach...
<Turl> arokux2: I do the same, and I have another filter to autotag emails that contain sunxi or allwinner, so they go green on my MUA :)
<arokux2> Turl, was thinking to do the same now :)
<arokux2> Turl, what is sunxi-devel branch for?
tinti has joined #linux-sunxi
<Turl> arokux2: so ping me when you mail your series and I'll add it to the tree
<arokux2> Turl, not sure it works as described. mripard does not add his in-review patches to this branch.
<Turl> arokux2: that's because I manage it, and last cycle I got lazy/confused when mripard sent a lot of series together and iterated on them :)
<Turl> arokux2: that's why I'm making a pledge to add them on the spot now :)
<arokux2> Turl, I see, will ping you.
focus has joined #linux-sunxi
\\Mr_C\\ has quit []
FDCX_ has quit [Ping timeout: 260 seconds]
hramrach has quit [Remote host closed the connection]
deasy has joined #linux-sunxi
vincenzo has joined #linux-sunxi
ykchavan has quit [Quit: Leaving]
<arokux2> Turl, how comes mainline works if there is no AXP driver? the AXP hardware is internally configured with some sane defaults?
<Turl> arokux2: uboot
hipboi has quit [Quit: Leaving]
<arokux2> Turl, you are damn right!
drachensun_ has quit [Ping timeout: 245 seconds]
bamvor_ has quit [Read error: Connection reset by peer]
tzafrir has quit [Ping timeout: 248 seconds]
uwe__ is now known as uwe_
<Tsvetan> oliv3r use always max settings 30mA as this way port will be fastest, FETs do not consume any current but if you set the port to be 30mA type it will rise and fall faster
<arokux2> Tsvetan, did you mean a hobbyist license for Eagle as the cheapest one?
<oliv3r> Tsvetan: yeah but in this case, it hardly matters, since the FET will only be turned on at the load of the module, and turned off at unloading
<Tsvetan> arokux2 I have not bought Eagle licensee recently so I dont know
<oliv3r> Tsvetan: it's a light-switch sort of, so do you still want it to be 30 mA then?
<Tsvetan> but there was optione for non commercial use which was eur 100
<oliv3r> there's even a shareware/free version iirc
<Tsvetan> oliv3r if the port is set to lower current it will be slower
<oliv3r> Tsvetan: you should replace/add the i2c eeprom with a SPI-nor one
<oliv3r> Tsvetan: but that's okay, we only turn it on once and off once
<Tsvetan> sure it will make no difference for this FET drive
<oliv3r> Tsvetan: a10/a20 can boot from SPI, so if we put u-boot into spi-nor, we can boot from USB, sata without mmc :)
<Tsvetan> oliv3r I though about this
<Tsvetan> is 8MB enough for uboot
<Tsvetan> or we need 16MB
<oliv3r> 256 kB i think
<oliv3r> maybe 512 kB to be safe
<Tsvetan> Ive seen miracast dongle design from rockchip
<Tsvetan> they use 16MB Flash and 256MB RAM
<oliv3r> that's small
<Tsvetan> to implement the miracast software
<Tsvetan> they say they run linux not android
<oliv3r> but shouldn't be to hard with any of those android tv sticks we have allready
<oliv3r> big problem with rockchip, is closed bootloader
<oliv3r> i think they use u-boot; but no sources hve been released
<oliv3r> yeah android might be tight with 256 MB ram and 16 MB Flash
<WarheadsSE> of course no
<oliv3r> but 512 kB SPI-nor should be more then enough to boot from other medium
<Tsvetan> so
<oliv3r> 8 MB would be enough for u-boot + small kernel + small initramfs
<Tsvetan> if we put 8MB SPI Flash on A20 + HDD
<oliv3r> but the idea is to have u-boot in a small SPI-nor, and then it can boot from usb, sata, nand etc
<Tsvetan> it will be able to boot?
<oliv3r> Tsvetan: you can do anything you want
<oliv3r> in theory, yes
<oliv3r> arokux2 did usb for u-boot
<oliv3r> i'm going to test the mainline ahci driver today/tomorrow
deasy has quit [Remote host closed the connection]
<oliv3r> and then see if we can integrate ahci support into u-boot too
<WarheadsSE> that would be nice
<Tsvetan> we can make small patch board with SPI Flash and few jumper wires
<Tsvetan> for experiments
<oliv3r> for that, 256 (tight, but possible) or 512 kilobyte NOR should be eough
<oliv3r> Tsvetan: oh! that's a good idea too
<oliv3r> i don't kno which SPI port BROM boots from
<Tsvetan> SPI1
<Tsvetan> SPI0 is connected to the SD card
<Tsvetan> IIRC
<Tsvetan> there is page with the boot sequence in A20 datasheet
<Tsvetan> you say 512kb but this will not hold the complete uboot? I think in SD card we make 16MB partitin for it
<oliv3r> yeah i think we ported that to the wiki
<oliv3r> Tsvetan: SPL is 24k; BROM will only boot 24k from its media's
<oliv3r> SPL can then chainload u-boot from SPI, which is 224 i think currently
<oliv3r> hno: ^
<arokux2> guys, I'm thinking.. could we have some donate buttons?
<Tsvetan> ok tell me what amount of Flash you need and we will prepare few SPI flash modules for A20-OLinuXino
<Tsvetan> to send you for experiments
<oliv3r> arokux2: you want people to donate to us?
<oliv3r> arokux2: dunno
<oliv3r> mnemoc: ^ your call
<oliv3r> Tsvetan: i'm goin gover to see which SPI is used
<oliv3r> Tsvetan: but frankly we have no clue which SPI flash does or does not ork :)
<oliv3r> Tsvetan: but if it's a 8 pin DIP socket, any chip can be put in there
<Tsvetan> isnt it better to have 8MB to may put more stuff if necessary
<Tsvetan> even small openWRT
<oliv3r> yeah
<oliv3r> you should be able to put bootloader + kernel +initramfs in 8mb
<oliv3r> but isn't 8mb MUCH more expensive?
<oliv3r> well for testing anything goes really :)
<oliv3r> and if its DIP you can swap it out :)
<Tsvetan> for development it really doesnt matter :)
<oliv3r> just gotta figure out from what SPI port it boots
<oliv3r> i'll go read boot0 SPI-nor code
atiti has quit [Ping timeout: 260 seconds]
<oliv3r> crap looks like the SPI nor stuff as dumped from boot0
<oliv3r> and boot1 code drops :S
<oliv3r> but doesn't matter hugely, gotta figure out how BROM loads data from SPInor mostly anyway
<oliv3r> and the assembly doesn't look too horrid
<oliv3r> Tsvetan: SPI0
<oliv3r> Internal BROM
<oliv3r> ?
<oliv3r> Support system boot from NAND Flash, SPI Nor Flash (SPI0), SD Card/TF card (SDC0/2)
<oliv3r> a10 says 'dedicated SPI controller for boot'
<oliv3r> also SPI0 on a10
<oliv3r> Tsvetan: SPI0 is mapped to two sets of pins, either PortC (nand) so that's not wise
<oliv3r> OR
<oliv3r> Port PI
<oliv3r> and the cool thing about PI, is that it is shared with UART5 an UART6
<oliv3r> and i think UART5 and UART6 on the olimexino are just plain extension ports
\\Mr_C\\ has joined #linux-sunxi
<oliv3r> meaning a simple extension connector -> spi should be enough
<atsampson> so does it try both PortC and PortI on startup, or only one of them?
massi has quit [Quit: Sto andando via]
<arokux2> oliv3r, (on donating) why not? we good very a great job here.
<arokux2> we could pay our hosting and buy some hardware, if needed
<oliv3r> Tsvetan: GPIO-2 and EXT1 have uart5 and 6, so a little annoying to get pins out, it certainly is possible
<oliv3r> arokux2: we get hardware donated :)
<oliv3r> arokux2: you shouldn't do this just to get donated/payed :)
<arokux2> oliv3r, shouldn't do what?
<oliv3r> arokux2: and we have only done a few little things out of passion ;)
<oliv3r> arokux2: write great code!
<oliv3r> i think hosting is being donated to us too
<arokux2> oliv3r, donations are not bad, especially if you see your work is actually used by busynesses to make real money.
<oliv3r> yeah business should pay :p
<oliv3r> arokux2: have you figured out if we need ohci or if that's included allready?
<arokux2> oliv3r, all those ppl with their allwinner tablets, maybe they are willing to throw a dollar or two at, why should we prohibit them from doing so?
<arokux2> oliv3r, no, I haven't. I have no USB keyboard
<oliv3r> what does lsusb say?
<arokux2> oliv3r, about what?
<oliv3r> hubs
<oliv3r> it usually lists usb1.1 and usb2.0 hubs
<arokux2> oliv3r,
<arokux2> oliv3r, this is with mouse connected.
<oliv3r> so cool seing the wlan adapter there too
<oliv3r> Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
<arokux2> oliv3r, i'm not going to investigate where ohci is needed. I have no usable USB 1.1 hardware
<oliv3r> there, we deff. have a 2.0 hub
<arokux2> whether*
<arokux2> oliv3r, well, EHCI is 2.0
<oliv3r> arokux2: IF USB 1.1 IS required (and ehci really only does offer 2.0 support) it may be required eventually; as a lot of devices are still 'only' usb 1.1 :(
<oliv3r> the strange thing is, the mouse works
<arokux2> oliv3r, I'll add it if demand appears, not "just to add it"
<oliv3r> so it's quite possible that this hardware does both transparantly
<arokux2> oliv3r, I'll see how EHCI patches will pass review etc.
<oliv3r> but it's also possible your mouse is a usb 2.0 capable one
<arokux2> oliv3r, yep
<oliv3r> but i only see usb 2.0 hubs
<oliv3r> i'll test your patches very soon
<oliv3r> i got my git tree sorted, so i can switch easily again
<arokux2> Logitech USB Optical Mouse
<oliv3r> i'll pull turls patches and add yours
<arokux2> you do not see this one?
<oliv3r> yeah that's hy i say, it's posisble your mouse is 2.0 copatible
<oliv3r> i have a lot of usb 1.1 hardware connected to my server for example, USB carddreader (1.1) USB printer (1.1) USB UPS (1.1)
<arokux2> oliv3r, patches welcome
<oliv3r> i noticed that 1.1 was needed, because when i compiled only EHCI driver, without OHCI driver, half of the shit stopped working :)
<arokux2> oliv3r, my next goal is AXP
<oliv3r> but it's quite supprising, since the original code doesn't do ohci differently either
<oliv3r> unless it's fully backwards cmpatible
<oliv3r> arokux2: yeah axp is cool; but quite complex :(
<arokux2> not really
<arokux2> quite dangerous I'd say :D
<mnemoc> oliv3r: I don't think we need donations (as community) at the moment, but I assume there are a few developers who would really appreciate to get some payed sunxi-related work
<oliv3r> someone on ML is asking for it :)
<mnemoc> oliv3r: so maybe a wiki page about hirable community-endorsed devs would be a good idea
<arokux2> mnemoc, you mean we do no have something we could spend money on?
<mnemoc> i can't think of anything beside paying devs on need to dedicate more time on the project
<mnemoc> oliv3r and turl are always whining about how poor they are for example
<arokux2> mnemoc, maybe we just pile up the money and once we need something we use it.
<mnemoc> you mean opening a paypal account for.... and just collect money there? it might be complicated considering taxing and transparency. but I have nothing against it
<mnemoc> anyone here living in .ch? :p
<arokux2> why .ch?
<mnemoc> taxing paradise :p
<arokux2> leviathanch, is from ch
<mnemoc> this is a community, let the community decide
eebrah_ has joined #linux-sunxi
<mnemoc> but for server, builder and domains we (atm) don't need money
<mnemoc> i would prefer to get turl, oliv3r and others a good sunxi-related job :p
<arokux2> mnemoc, who is paying for that?
<mnemoc> basically me
<arokux2> then I do not understand why you don't want to get it paid by donations?
<arokux2> to much work to set them up?
<mnemoc> i enjoy been here. i don't expect to get money for having fun. and I don't really need it.... so it's like hobby-budget :p
<mnemoc> it's not that expensive
<mnemoc> and i have a decent formal income source
<mnemoc> also, others devs do much more for sunxi than I
<mnemoc> that's why I think donation in the form of hiring community-endorsed devs to do sunxi-related dev is much better, and far more transparent
tinti has quit [Ping timeout: 248 seconds]
<mnemoc> and we can just make a (restricted) wiki page for that with a list of hirable people, resume-summary and contact info
<mnemoc> many projects do that
<arokux2> mnemoc, I'd like to get 20 hours per week job on sunxi related stuff too
<arokux2> I'm just thinking - all those ppl with their allwinner tablets, maybe they are willing to throw a dollar or two at, why should we prohibit them from doing so?
eebrah_ has quit [Read error: Connection reset by peer]
<Turl> maybe a bounty system? just an idea
<arokux2> was thinking about that too
<arokux2> actually today I had more general idea, not related
eebrah|noPC has joined #linux-sunxi
<arokux2> but... already there.
atiti has joined #linux-sunxi
<Turl> :)
tinti has joined #linux-sunxi
<mnemoc> I can also just setup a donations paypal account to pile up money for eventually helping someone to go to an event
<mnemoc> but choosing the blessed person might be difficult
<mnemoc> i doubt we can pay Turl's ticket to fosdem, but maybe oliv3r's bus or arokux2's train :p
<arokux2> we can do it by chance first
eebrah|noPC has quit [Read error: Connection reset by peer]
<arokux2> next time by chance from the rest :)
<mnemoc> donation@ donations@ or donate@ ?
<arokux2> doesn't matter
<mripard> also, if you're interested, the linux foundation has some budget allocated to the speakers
<mnemoc> please vote for the mail account name
<mripard> so if you want to get to a conference organized by the LF, you just have to give a talk basically :)
<mripard> (just to motivate a bit more to contribute to mainline :))
<arokux2> donate@
<mnemoc> mripard: ^^
<mnemoc> mripard: consider 3.10 is LTS, also LTSI, and likely to be next android's. what about maintaining a backported-mainline sunxi-3.10?
<mnemoc> DTS based obviusly
tinti has quit [Ping timeout: 256 seconds]
<mnemoc> it should also be a simple stepping stone toward linux-next
<mripard> yeah, that could make sense
<mripard> but backporting can be time consuming
<mripard> I can help, but I won't do it myself
<mnemoc> it can also not be backported, but the place to develop for mainlining
<mnemoc> 3.10 is pretty similar to master
<mnemoc> and we can make more commits, while preparing something mainline will accept
<arokux2> mnemoc, I'm not so much interested in this unless someone steps up and throws some cash
<mripard> well, much closer to master than 3.4, for sure
<mripard> but still quite different
<mripard> we're working on 3.13 right now
vinifr has joined #linux-sunxi
<arokux2> because all those chinese companies will use our 3.10 work (in there androids) and make money, I do not like it.
<arokux2> for me, it is fun to hack, but to hack on mainline
<mnemoc> is developing a decent sata driver for 3.10 that hard to submit to 3.13 once decent?
<arokux2> and not care about somebody who wants to sell their tablets
* mnemoc doesn't mind with people earning money thanks to open source software. and reducing fragmentation increases the chances of contributions
<mnemoc> and at the end, everyone wins
<mnemoc> if AW develops their own junk on 3.10, we lose. if they use our sunxi-3.10. we win.
eebrah|noPC has joined #linux-sunxi
<mnemoc> teaching fexc to generate .dts out of their .fex, they can even use DTS
<arokux2> mnemoc, maybe I'm selfish but I do not see it as a win if allwinner uses sunxi-3.10.
<mripard> mnemoc: on a general basis, yep, it's quite close
<arokux2> mnemoc, actually there is very little in fex what we normally put in dts
<mripard> yet, there's always a few boilerplate functions added here and there, some functions that are slightly different ,etc.
<arokux2> mnemoc, fex contains only settings - like module parameters in mainline
eebrah_ has joined #linux-sunxi
<mripard> and when backporting, you have to track all of that, and modify the drivers accordingly
<mripard> (or pulling all the other patches)
<mnemoc> as more drivers get dts-ised, the .dts will grow
eebrah|noPC has quit [Read error: Connection reset by peer]
<mnemoc> at least to enable the corresponding platform_devices
<mnemoc> and their pin muxing
<mripard> and which driver instances are loaded as well
<arokux2> mnemoc, allwinner should step up by themselves and offer something in return
<mripard> so I guess it could be doable
<arokux2> and not we should create 3.10 and beg them to use it.
eebrah_ has quit [Read error: Connection reset by peer]
tinti has joined #linux-sunxi
<mripard> arokux2: well, if you submit some code, everyone's free to use it
<mnemoc> arokux2: if sunxi-3.10 is future/mainline friendly, we won't need to beg
<mripard> that's the point of FOSS
eebrah|noPC has joined #linux-sunxi
<mnemoc> it's less work for them to join us (and share docu!) than porint script.bin upon 3.10
<arokux2> mnemoc, what about they share docu, we backport to 3.10? this way around I'm willing to help to.
<mnemoc> mripard has had contact with aw devs, I'm always blocked by eva
<mnemoc> after all, mripard is the official sunxi maintainer
<mnemoc> funnily one of the most open armv7-a platforms at the moment :p
<arokux2> mripard, I'm fine with everybody else using my code, I just do not want to put an extra effort to make LTS more usable by ppl ignoring us (at least I have this impression)
<mnemoc> arokux2: think in normal fellows wanting to run android 5 in their sunxi devices
<mnemoc> not only in the manufacturers
<arokux2> mnemoc, then fellows need to jump in and submit patches
<mripard> mnemoc: you can run android5 on any kernel you want
<mripard> as long as there is the proper kernel APIs
<mripard> namely, the only one missing for now is ION
<mripard> I'm sorry, I have to pack for my flight
eebrah|noPC has quit [Ping timeout: 252 seconds]
<mnemoc> G will manage to add something in android 5 that will not work on top of an androidized 3.4
<arokux2> mripard, nice flight!
<mripard> but we can continue this discussion later :)
<mnemoc> *G*
<mripard> mnemoc: it's not part of their plan for now
<mripard> they are pretty much trying to do the opposite
<mnemoc> mripard: great
<mripard> do as much as they can to put everything into mainline
<arokux2> mnemoc, so unless Allwinner steps up *first* I'm not going to make their lives easier.
<mripard> anyway, bye, see you on your side of the atlantic :)
<arokux2> mripard, c u
<arokux2> mnemoc, btw, by holding that line we could actually press them to get docs
<arokux2> mnemoc, why would they want to give us docs, if sunxi-3.10 works already for them? they just take and use it.
<mnemoc> it will not work for their next SoCs
<mnemoc> so they might contribute patches
<mnemoc> or hints about how to solve mistakes
<mnemoc> they are already sending patches upstream!
<mnemoc> no other chinese vendor has done that
<arokux2> mnemoc, who? what patches?!
<mnemoc> mripard knows better, but he needs to pack
<arokux2> :)
<mnemoc> search for in lkml
<mnemoc> it's an awesome sign
<rm> anyone has Mele A2000G (A10 1GB)?
<arokux2> mnemoc, I cannot find anything
<arokux2> rm, A1000
<arokux2> mnemoc, are you sure?
eebrah|noPC has joined #linux-sunxi
<arokux2> mnemoc, all I find are strings from copyright
<mnemoc> yes, I saw posts from aw devs
<mnemoc> maybe it was in the linux-arm list
ykchavan has joined #linux-sunxi
<mnemoc> f* paypal doesn
<mnemoc> 't let me choose english if I pick .de as country
<arokux2> mnemoc, why are you blocked by Eva, btw?
hramrach has joined #linux-sunxi
tinti has quit [Ping timeout: 268 seconds]
<mnemoc> arokux2: for asking for gpl code too many times before our community became relevant
<arokux2> mnemoc, :)
<mnemoc> mripard otoh caught their attention by mainlining the initial sunxi related code
<mnemoc> instead of by annoying eva
<arokux2> mnemoc, I think we are relevant now and we should play smart not to lose over status
<arokux2> our* status
<mnemoc> that's another point in favour of sunxi-3.10
<mnemoc> if they can (ab)use community code, community becomes even more relevant
<mnemoc> and so they (can) become more open to talk and work with us
<arokux2> mnemoc, what if they take our sunxi-3.10 and then hack on it in house?
<mnemoc> sh*t happens
tinti has joined #linux-sunxi
<mnemoc> but when the next source drop appears, it will be closer to our code
<mnemoc> than if they start their own 3.10
<mnemoc> so, we win (less but win) anyway
<arokux2> mnemoc, next SoC based on 3.10 will appear in several years, to much time to wait
<mnemoc> it seems that paypal likes to screw open source projects that receive donations
<mnemoc> arokux2: google is actively working on an android-3.10 kernel branch
<mnemoc> arokux2: that kind of implies the next android will use it
<mnemoc> and aw is android-driven
_BJFreeman has joined #linux-sunxi
<arokux2> mnemoc, next aw SoC that is in development now probably will stick to the old android
<arokux2> mnemoc, so android5 will be used by the SoC that comes after it..
<mnemoc> yes, as A20/A31 were initiallz 3.3 based, and now 3.4
<arokux2> that is a year from now.
_BJFreeman is now known as BJfreeman
<mnemoc> so the A23, A40, and A51 or however they call their next generation SoCs will likely to have an initial 3.4 SDK and then get upgraded to 3.10
<arokux2> mnemoc, we better put allwinners in position where they are interested in putting docs public to encourage to show them some love and work on 3.10.
<arokux2> mnemoc, if we then work on 3.10 and they like it, they should publish more docs
<arokux2> mnemoc, it should be a game played by both partners. not us caring for them and they decide whether they could use us.
eebrah|noPC has quit [Ping timeout: 252 seconds]
<mnemoc> i'm just an open source (mostly licensing stuff BSD or MIT actually) developer, not a business/sales/politian/strategies man
<arokux2> mnemoc, but you like to have docs for your hacking right?
<mnemoc> for me, the more people uses our tree, the better
WarheadsSE has quit [Ping timeout: 240 seconds]
<mnemoc> arokux2: we have *received* (partial) user manual of their latests SoCs. the A10's was leaked.
<mnemoc> that is already progress
<arokux2> mnemoc, from Allwinner directly?
<mnemoc> yes
<mnemoc> they are limited, but much better than nothing
WarheadsSE has joined #linux-sunxi
<arokux2> mnemoc, they've sent an e-mail to our ml?
<mnemoc> no, given to mripard and authorized him to share them
drachensun has joined #linux-sunxi
<arokux2> mnemoc, good. we did some job for them in return. but they should keep giving us more "fuel".
eebrah|noPC has joined #linux-sunxi
<mnemoc> greed makes harder to enjoy the little successes
<arokux2> true.
<arokux2> mnemoc, I'm talking with sunxi-3.10 in mind. why not make them an offer like this and ask for docs in return?
<arokux2> mnemoc, as I understand it, we have everything we need for current SoCs to mainline them, don't we?
<arokux2> mnemoc, except of manpower :)
<arokux2> mnemoc, if we force Allwinner to become more open, the others should follow. I mean RK, and FOSS will win
vinifr has quit [Ping timeout: 248 seconds]
<arokux2> hno, ping
eebrah|noPC has quit [Read error: Connection reset by peer]
eebrah|noPC has joined #linux-sunxi
<arokux2> jukivili, ping
BJfreeman has quit [Ping timeout: 240 seconds]
FDCX_ has joined #linux-sunxi
eebrah has joined #linux-sunxi
tinti has quit [Ping timeout: 248 seconds]
<ykchavan> I am wondering if Cubieboard2 and CubieTruck are both A20+mali with small hardware component differences, will their sdcard/nand images will work with each other?
<arokux2> any idea for fixing this?
<mnemoc> the only really important detail when using images is to use the right script.bin
<mnemoc> arokux2: sorry, f* sucks badly.
<mnemoc> arokux2: there are several parts of the SoCs we don't have any driver at all, or incomplete info
<arokux2> mnemoc, what are they?
<mnemoc> arokux2: but we can get a usable mainline for all of them having enough people cleaning/unifiying/researching
<mnemoc> arokux2: a cheap mobile carrier
<mnemoc> arokux2: who left me (again) like an hour offline
<arokux2> mnemoc, I'm not asking about blau, about unknown parts of the SoCs :)
<mnemoc> can and gps are the first to come to my mind
<mnemoc> not even talking about the GPUs and VPU
<mnemoc> no ps/2 driver either
<arokux2> GPU is arm
<mnemoc> or pvr
<mnemoc> so out of scope. and RE-only
<arokux2> GPUs won't be given to us anyway. it is not allwinner jurisdiction
<mnemoc> VPU is been nicely REed too, but they COULD open source it
<mnemoc> it's likely that the blanks in the user manual also correspond to 3rd party IPs
<mnemoc> but basic "hints" can allow mainlineable drivers
<mnemoc> or extend others, like happened with serial and usb
<arokux2> mnemoc, what usb hints do you mean?
<mnemoc> arokux2: it was discovered that the usb IP is based on the already mainlined musb IP
akaizen has quit [Remote host closed the connection]
<mnemoc> arokux2: so we don't need to mainline the whole monster, only extend musb a bit
<arokux2> mnemoc, ah, OTG. yes, but that is obvious from the code anyway
akaizen has joined #linux-sunxi
awafaa has quit [Ping timeout: 240 seconds]
wingrime has joined #linux-sunxi
<wingrime> Turl: are you added sunxi_clock_protect(), but how then we will deal with PM suspend?
akaizen has quit [Remote host closed the connection]
<arokux2> shit. USB support isn't as good as I thought.
<arokux2> I'm trying to get wifi working, but something is wrong with usb.
ssvb has quit [Quit: Leaving]
<ojn> Turl: You don't have to cc me on every patch, Maxime is the sunxi maintainer, not me.
tinti has joined #linux-sunxi
tinti has quit [Excess Flood]
iamfrankenstein has quit [Ping timeout: 264 seconds]
tinti has joined #linux-sunxi
ssvb has joined #linux-sunxi
<wingrime> Turl: ddr can mbus can be protected as well in pm driver
wingrime has quit [Read error: Connection reset by peer]
wingrime has joined #linux-sunxi
eebrah has quit [Ping timeout: 245 seconds]
<oliv3r> mnemoc: we have mainline sata driver; it just needs to get properly tested and mainlined :)
<mnemoc> oliv3r: that's also one of the points for sunxi-3.10. freedom development *toward* mainline but not rebased
<mnemoc> we can move much more stuff to a DTSed 3.10 than to mainline
<mnemoc> and the evolution of the cleaning gets proper history too
<mnemoc> that also means more testing
<mnemoc> because you can get a more usable kernel sooner
<mnemoc> (due to quality issues)
<arokux2> mnemoc> we can move much more stuff to a DTSed 3.10 than to mainline| why?
<mnemoc> because mainline has VERY strict quality requirements and merging windows
<mnemoc> a DTSed sunxi-3.10 is controlled by us
<mnemoc> so we CAN get the current drivers there, DTSize them, and port them to the common infrastructure incrementally
<mnemoc> eventually reaching mainline-ability
<mnemoc> but been TESTED in real life already by a larger user base
<arokux2> then you better select 3.11
<mnemoc> 3.10 is LTS, LTSI, and likely andriod 5's
<mnemoc> 3.11 won't last long
<mnemoc> 3.100 will
<mnemoc> 3.10
<arokux2> sorry, I do not see this as advantage. you can keep posting your patches to mainline and improve them until they are taken
<mnemoc> on lkml normal people won't use them, only a few devs
<mnemoc> on a more complete branch (even if drivers are still crappy) normal people can test and run it in real life
<arokux2> mnemoc, 3.4 is for real life
<mnemoc> 3.10 can be real life for DTS
<mnemoc> while 3.4 is real life for script.bin
<mnemoc> the distance betwen 3.4 and mainline is huge
<mnemoc> 3.10 is post-multiplatform and already has sunxi support
<mnemoc> so the distance with mainline is shorter
<mnemoc> a real life stepping stone
<mnemoc> 3.4 is just a legacy real life
<mnemoc> stable
<arokux2> mnemoc, I do not think we have resources to mainline twice
<mnemoc> it's not mainlining twice
<mnemoc> it's doing the mainlining in a real-life usable git branch
<mnemoc> instead of private repos and mailing lists vN patchsets
<mnemoc> jumping from a already-mainline-quality 3.10 driver to mainline is much simpler
<mnemoc> and it's REALLY tested already
<arokux2> mnemoc, if ppl want to use our kernel they'd need to take it from our kernel trees and compile it anyway.
<arokux2> mnemoc, so they could checkout any kernel from our trees. it shouldn't be 3.10
<mnemoc> 3.10 allows us to have single bin for all sun[4567]i
<arokux2> mnemoc, and how current mainline prohibits this?
<mnemoc> 3.10, as 3.4, is long term maintained, other non-sunxi related fixes will get backported, also android stuff
<mnemoc> mainline moves
<wingrime> mnemoc: you have time to do it?
<mnemoc> fine for devs, not for adventurous normal users
<mnemoc> wingrime: no, i'm just advocating
<arokux2> mnemoc, android is an advantage for allwinners, not for us.
<mnemoc> wingrime: btw, AXP = Allwinner's X-Powers
<mnemoc> arokux2: android is advantage for users
<rm> just forget about Android, who cares about it
<wingrime> mnemoc: I not agree with axp's remorval
<rm> I see the fact that 3.10 is a long-term kernel as a huge advantage alone
<arokux2> mnemoc, why? they have already android shipped
<mnemoc> arokux2: crappy and they can't add drivers
<mnemoc> also 3.10 is better for linux users as it's LTS
<mnemoc> note the diff between a *user* and a *dev*
<arokux2> mnemoc, they should ask for sources and not relying on us doing all the work
<rm> long term are 3.2, 3.4 and 3.10
<rm> with the current pace the next long term will be 3.16 :)
<arokux2> by 3.16 we will hopefully mainlined everything :)
<mnemoc> sunxi-3.10 can help us to be (user) usable in mainline 3.16
<mnemoc> as it can move much faster than ML-submited/quality controlled submitions
<arokux2> mnemoc, I see sunxi-3.10 as free labor for allwinner who do (almost) nothing to help. they will clearly profit from it so we should force then to give us smth in return.
<arokux2> force them*
<arokux2> living our lives without considering this is not wise. we need docs.
<rm> you say that like allwinner cares about you
<rm> or about linux-sunxi in general
<mnemoc> arokux2: by asking I got banned. by giving mripard got support
<rm> if android 5 is 3.10, they will just go on develop their own horrible code to get 3.10
<mnemoc> I prefer allwinner using our sunxi-3.10 as base and not their own horrible thing
<arokux2> rm, they could well do it, but they also can cherry pick our work, done for free for them.
<mnemoc> then the SDKs will give us more useful diffs
<mnemoc> arokux2: not from 3.4, not from mainline.
<arokux2> mnemoc, I'm interested to know what mripard has actually got from them.
<mnemoc> they will start from google's android-3.10. or hpefully our (androidized) sunxi-3.4
<rm> personally I love that 3.10 has been declared long-term
<rm> I am tired of upgrading to 3.7/3.8/3.9 only to find it pulled from under me
ZetaNeta has joined #linux-sunxi
<mnemoc> err, androidizd sunxi-3.10
<arokux2> mnemoc, if it is something we could read in code like addresses of clocks we got ZERO
<rm> with the new latest-greatest release pushed as the new stable
<arokux2> we need cedar, gps, etc.
<wingrime> mnemoc: ask ben about git history
<mnemoc> wingrime: benn doesn't work for aw, he works for cubietech. and he gets the same SDKs as olimex
popolon has joined #linux-sunxi
<mnemoc> wingrime: the diff is that he has friends inside who can take the time to build custom blobs for him
<mnemoc> wingrime: or answer not-NDAed questions, in chinese
<mnemoc> but they can't give him the git repo
<wingrime> mnemoc: .repo folder
<mnemoc> *g*
<mnemoc> current SDKs come with mutilated .repo
<mnemoc> history less and broken links
<wingrime> mnemoc: we have git in SDK, but better this with .repo dir
<arokux2> mnemoc, you see they've learned from the leak and closed tighter
mcbrick has quit [Ping timeout: 260 seconds]
<mnemoc> arokux2: please stop wasting time thinking in aw as an enemy. and how to attack back or negotiate. and remember this is about open source, and about GIVING. not GETTING.
<wingrime> mnemoc: yes, I think aw help as as they cam
<arokux2> mnemoc, to give we need docs, don't we?
<wingrime> mnemoc: but thats non offical way
<mnemoc> we can do a complete usable 3.10 without anything beyond what we already have
<wingrime> mnemoc: they also have 3rd NDA agrement with IP vendors
<mnemoc> not mainline-quality, but an stepping stone to become mainlineable
<arokux2> mnemoc, open source isn't about GIVING it is a way to develop software. it has nothing to do with sacrifice.
<rm> it is GPL anyway, if it was BSD they could close (in a way, "truly steal") our improvements
<mnemoc> sure it's a way of development, but it doesn't assume EVERYONE will contribute back
<mnemoc> rm: yes, thankfully the kernel is GPL.
<mnemoc> rm: but in practice GPL is not a warrnaty that any vendor will actually provide the stuff unless they WANT to
<mnemoc> even big western vendors take ages to release GPL stuff
<mnemoc> so we need to convince them to join us as friends, not to attack them
<mnemoc> slowly they fear less to provide more info
vinifr has joined #linux-sunxi
<mnemoc> our stuff is even more open than most TI's or broadcom's or fsl's
<arokux2> mnemoc, who says about attacking? we cannot attack them anyway.
<wingrime> arokux2: we have much unoffical support from aw
<arokux2> mnemoc, we offer them sunxi-3.10 and see what they say.
<mnemoc> the best way to offer them sunxi-3.10 is to make sunxi-3.10
<mnemoc> and let them know what we need to go farther
<arokux2> mnemoc, and what we need? to put everything from there code drops into 3.10 will take a year..
<arokux2> their*
<mnemoc> first 3.4 and unification. then 3.10 and DTSation and common post-multiplatform frameworks. then mainline!
<mnemoc> sadly I don't have the time to do it myself :(
<arokux2> by that time mainline will *change* and diverge from 3.10
<mnemoc> not that much
<mnemoc> multi-platform was the biggest thing, and that is already there (together with the sunxi core) in 3.10
<mnemoc> and I
<mnemoc> and I'm not talking about bulk 3.10->mainline, but driver by driver
<mnemoc> as soon as they get mainlinable quality
<mnemoc> but a 3.10 with crappy drivers allows real life testing of the code that will be mainlined
<mnemoc> but real users, not only advanced devs
<wingrime> mnemoc: whait usb and sdio mainlined
<wingrime> *wait
<arokux2> as said: real users would need to compile the 3.10 from our private branches anyway. what is the difference if the compile mainline or 3.10?
<jelly-home> mainline can often be buggy in unrelated ways
<arokux2> jelly-home, then we'll get testers for mainline :)
<jelly-home> I'd rather have separate bug hunting for $arch and for mainline
<mnemoc> sunxi-3.10 can remove the need of private branches for everything
pepee has joined #linux-sunxi
<mnemoc> and allow real users to just use daily prebuilts, etc
<mnemoc> multiplatform
<mnemoc> 3.10 is a safe base
tinti has quit [Ping timeout: 264 seconds]
<mnemoc> and been LTS it will "receive" all sort of fixes
<mnemoc> helping our users
<mnemoc> and we will only need to care about getting sunxi drivers into mainlineability
<mnemoc> and keeping history of the process in a central location
<pepee> hi. are there disk images to test the latest cedarx open source drivers?
<pepee> hmm, I think I found some... do you think the lubuntu images for the cubietruck would work in MK802 II?
tinti has joined #linux-sunxi
tinti has quit [Remote host closed the connection]
ykchavan has quit [Quit: Leaving]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
awafaa has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
arokux2 has quit [Quit: Leaving]
FDCX_ has quit [Ping timeout: 260 seconds]
iamfrankenstein has quit [Ping timeout: 268 seconds]
deasy has joined #linux-sunxi
arokux2 has joined #linux-sunxi
arokux2 has quit [Client Quit]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
iamfrankenstein has joined #linux-sunxi
tzafrir has joined #linux-sunxi
drachensun has quit [Read error: Operation timed out]
iamfrankenstein has quit [Ping timeout: 268 seconds]
drachensun has joined #linux-sunxi
arokux2 has joined #linux-sunxi
granite_crusher has joined #linux-sunxi
<granite_crusher> arokux2: do you still need ehci tested on A13?
<arokux2> granite_crusher, yes.. but not at the moment. have you booted mainline already once?
<granite_crusher> still not
<arokux2> granite_crusher, the situation is the following. A13 needs some work to be done. I do not have hardware, so you need to be patient and test it more then once...
<arokux2> granite_crusher, so maybe you try to boot mainline, just to learn how to do it?
<granite_crusher> ok
<arokux2> granite_crusher, here is how:
<granite_crusher> is there difference in hardness depending of mainline branch? 3.12 or 3.13?
<granite_crusher> or devel?
<arokux2> granite_crusher, no, no difference. but you better go with: git clone git:// -b sunxi-next
<arokux2> granite_crusher, however if you have kernel tree already you do not need to do full clone and waste your HDD space
<arokux2> granite_crusher, if you have a kernel tree, you could just add to it. ask me how if you want to do it this way.
<granite_crusher> well I have old 3.4
<granite_crusher> I wouldn't want to get it touched, because it worked
<granite_crusher> but probably I could update u-boot?
<arokux2> granite_crusher, it won't do any harm to your old 3.4
<arokux2> granite_crusher, you'll just add a separate branch
<granite_crusher> branching in my laptop - sounds cool
<granite_crusher> :)
<arokux2> granite_crusher, u-boot.. yes, I think it is better to update it. there were some changes, not sure you need them, but just to be on the safe side.
<arokux2> granite_crusher, so decide what you'll do and ask how to proceed if you do not know yourself
eebrah|noPC has quit [Ping timeout: 260 seconds]
<arokux2> granite_crusher, and ping me, so that I do not need to check my irc client.
<granite_crusher> arokux2: this probably will sound silly, but what initramfs exactly is?
<granite_crusher> I didnt need that when compiling kernel 3.4
<arokux2> granite_crusher, there is no storage driver in mainline yet. so you either configure network boot - rootfs will be on different computer, or you build rootfs into initramfs. initramfs gets built into kernel image. it is like rootfs but in RAM, hence the name.
<granite_crusher> oh that explains a lot
<granite_crusher> arokux2: so I could simplly use that minimal rootfs I generated for 3.4?
<arokux2> granite_crusher, yep. the smaller the better - loads faster. it should be in cpio format - at least it worked for me.
<granite_crusher> arokux2: network boot is impossible without usb (using usb dongle on olimex)
<granite_crusher> cpio?
<arokux2> granite_crusher, no ethernet?
<granite_crusher> no
<arokux2> yes, cpio...
<arokux2> granite_crusher, which one do you have?
<arokux2> and what minimal rootfs have you generated?
<granite_crusher> well, I guess it is not so small, I used bootstrap in debian
<arokux2> oh
<granite_crusher> hell.. I just checked, it is not minimal at all
<arokux2> you need buildroot
<arokux2> granite_crusher, i can give you one
<arokux2> granite_crusher, one thing.. there is some work to be done before you can test usb and i'd need to guide you. this will cost me time. and if you are going to disappear in between it is not worth it. so you'd need to do it till usb works in mainline for your board. agree?
<oliv3r> mnemoc: i agree, 3.10 LTS is the next big stepping stone
<oliv3r> mnemoc: can you setup the proper branches and i see if i can backport my stuff, shouldn't be a biggy
<granite_crusher> arokux2: you mean this night? or at all... cause else my fiance would kill me :)
<mnemoc> oliv3r: \o/
<oliv3r> 3.10 should become the abomination that is 3.4 + 3.15 sort of
<arokux2> granite_crusher, of course not this night :)
<mnemoc> oliv3r: androidized or not yet?
<arokux2> granite_crusher, I just were helping some ppl here and they disappeared in the middle of the process so that is why I'm asking
<mnemoc> oliv3r: I'll make a reference-3.10 now
<arokux2> mnemoc, :D
<oliv3r> mnemoc: track both LTS and goog'es?
<oliv3r> since we _need_ android support for the android users :)
<arokux2> mnemoc, now you have time :p
<oliv3r> which are probably more
<mnemoc> oliv3r: manually tracking
<mnemoc> oliv3r: it's what I merge into sunxi-3.x when doing jumps
<arokux2> oliv3r, mnemoc, you'd better set up mailing list and backup
<oliv3r> mnemoc: and we need to start moving people onto 3.4 harder, so 3.0 can die
<mnemoc> which is the latest android-3.x + latest stable v3.x.y
<oliv3r> so we then only have 3.4 and 3.10
<oliv3r> i don't know :(
<mnemoc> 3.4 as legacy (script.bin) for unification and aw code integration
<mnemoc> and 3.10 for DTSization and migration to common frameworks
<oliv3r> yeah
<oliv3r> i think that's a good plan
<oliv3r> 3.0 is allready missing out on patches
<oliv3r> i'm planning to test 3.4 on my tablet tomorrow
<oliv3r> and probably hackup the TS driver
<mnemoc> yes, 3.0 is implicilitly deprecated already
<mnemoc> specially because A20 and A31 code is 3.3/3.4 based
<mnemoc> and makes no sense to backport that to 3.0
<mnemoc> so 3.0 RIP
* oliv3r nods
<oliv3r> yeah
<oliv3r> actually; today, sep. 22 2013 is the official death of sunxi-3.0
<oliv3r> :)
<mnemoc> *g*
<granite_crusher> arokux2: I own FOSS a lots, (using linux GNU debian already 5 years) and this looks good oportunity to get involved in giving back
<arokux2> granite_crusher, yes, very good, since you'll learn a lot in addition
<granite_crusher> actually just after I started to play with this ARM board I started to understand how little I knew about it
<arokux2> :)
<mnemoc> arokux2: I have a little spare time, yes. but I prefer to use it in more "visible" stuff. like preparing sunxi-3.10 and merging outstanding patches from the ML
<mnemoc> and give some movement to the wheel
<arokux2> mnemoc, I agree ML isn't necessary. but please setup backup (and verify it works), although it is not visible it is actually life saving.
<mnemoc> will request the ftp account to hetzner now
<arokux2> mnemoc, just think what will happen if we lose cedarx RE wiki pages
<mnemoc> and backup the wiki daily there
<arokux2> mnemoc, cool. thanks.
<mnemoc> arokux2: but don't forget it's ext4 over raid10..... so data is not in such a risky possition
<arokux2> mnemoc, kind of pity the wiki isn't backed by git. we could just put it on github..
<arokux2> mnemoc, yes. true.
<mnemoc> someone should write a pymediawiki bot turning history into git :p
<mnemoc> it shouldn't be that complex
<wingrime> mnemoc: yeax, cedar regs are much important info,recived with much pain
<wingrime> mnemoc: I can get some regs for gps.ko, but without HW it useless anyway
<mnemoc> wingrime: how are your EE skills?
ZetaNeta has quit [Ping timeout: 268 seconds]
<mnemoc> I have an A10/BDD sodimm + kit here, which exposes all pins, and nothing to do with it
<wingrime> mnemoc: no IC
<mnemoc> wingrime: maybe benn can tell us what RF IC to try
<mnemoc> IF you can make a test board using this A10/BDD kit
<wingrime> mnemoc: .....
* wingrime still think who need this, if there ready-to go uart gps
<wingrime> mnemoc: I have a20 olimex this all pins exposed
<mnemoc> ok
<mnemoc> better
<wingrime> mnemoc: but not sure a20 have gps
<mnemoc> true
<wingrime> mnemoc: it realy mostly useless, all modern gps ic come with NMEA over uart
<arokux2> isnt A10 dumped anyway?
<mnemoc> no, they still sell it
<wingrime> arokux2: a10 fine for some usages
<mnemoc> and promised to sell it for at least another 3 years
<mnemoc> also, software is far more mature than for a20
<wingrime> mnemoc: linux-sunxi also can't shutdown on a20
<mnemoc> patch?
<wingrime> mnemoc: still have no idea
<wingrime> how fix
<wingrime> mnemoc: but I sended patch for sound resume pcm-audio (100% same with hsang one for hdmi-sound)
<wingrime> mnemoc: if you press *space* with mplayer to pause playback, you get mplayer hung, and big ooops
<mnemoc> :o
<wingrime> mnemoc: also, I find bug with hdmi backlight , when you not touch mouse and kdb for a long time, screen goes off, but on a20 It can't back to on again
<wingrime> mnemoc: but still no idea where
<wingrime> to fix
<arokux2> wingrime, is this bug present in original code?
n01 has quit [Ping timeout: 260 seconds]
<wingrime> arokux2: yes, cb2 internal firmware have such bug
<wingrime> arokux2: thats in nand
<wingrime> ssvb: are you tested new code for g2d from aw?
<granite_crusher> arokux2: before I go to sleep, I downloaded buildroot. It looks it have it's own toolchain... I cant figure out how to do a thing with it; also I found that I have separate 3.11.0-rc4 of kernel source last time I tried to get to mainline...
<ssvb> wingrime: only with 3.3 kernel, several months ago
<arokux2> granite_crusher, you'd need a config for buildroot. I can give you mine. I use linaro toolchain with it.
<arokux2> you can do git pull in this mainline tree
<wingrime> ssvb: I think patric patch need for g2d need spilt
<ssvb> wingrime: have you found some problems with g2d?
vinifr has quit [Quit: Saindo]
<wingrime> ssvb: no, his code style, incorret indent etc
<ssvb> wingrime: that's the original allwinner coding style, not patrick's :)
<granite_crusher> I have toolchain, and it looks I have to go. goodnight!
<wingrime> ssvb: thats not difficult to fix in patch (I not force him remake file)
granite_crusher has quit [Quit: Page closed]
<mnemoc> oliv3r: stage/sunxi-3.10 up. android-3.10 + 3.10.12 ..... multi_v7_defconfig builds :p
<mnemoc> oliv3r: all yours
<wingrime> ssvb: also, I don't think minal patchset for cedar 4 and 7 merge good idea
<wingrime> *minimal
<ssvb> wingrime: fixing the coding style for g2d is not difficult, but it will just make merging any potential future fixes from allwinner more problematic
<mnemoc> oliv3r: let's try to make sunxi-v3.10.12-r1 the backport of everything sunxi already mainlined
<wingrime> ssvb: good point, but, it can be resolved with eye
<wingrime> ssvb: about cedar, do you think better ask him about full merge patchest? not minimal strange one
<wingrime> ?
<mnemoc> oliv3r: a sunxi_defconfig for testing and nighly builds would be greatly appreciated =)
<ssvb> wingrime: I believe that patrick is on a vacation now (as he mentioned in his last messages)
<wingrime> mnemoc: we have some old defconfigs (not sunxi_ ) maybe remove them?
<wingrime> mnemoc: people have chance get problem when using them
<mnemoc> wingrime: in 3.10??
<mnemoc> i haven't merged anything sunxi there
<wingrime> mnemoc: no, in 3.4
<arokux2> AXP20x are all really revisions of the same thing, how do you think?
<mnemoc> wingrime: they all build for me.... that's how I "test" the patchsets atm
<mnemoc> wingrime: building all "our" defconfigs
<mnemoc> hansg studied the axp drivers in deep and has a good idea how to unify them, but no time to do so
<wingrime> mnemoc: than whats differences with a12_defconfig and sun5i_defconfig ?
<mnemoc> sun5i is a familz, a13 is a particular soc
<mnemoc> a12
iamfrankenstein has joined #linux-sunxi
<wingrime> can you make diff?
<arokux2> mnemoc, them? only 209 and 152 are used now
<wingrime> I not sure we need special a12 and a13 defconfig while we have sun5i_
<mnemoc> wingrime:
<mnemoc> arokux2: that's up to the board manufacturer
<arokux2> mnemoc, I see. I'm going to mainline 152 and 209 only as there is no known hardware with something else.
<mnemoc> a12 and a13 are pretty different.... same for a13 vs. a10s. not sure how safe is it to merge them into a sun5i_defconfig until we have properly integrated soc-detect and platform_devices based on script.bin
<wingrime> mnemoc: almost crazy diff
<mnemoc> wingrime: yes, they are a mess
<wingrime> mnemoc: logicless
<wingrime> mnemoc: but axp are detectable?
<mnemoc> there are a bunch of outstanding commits to standarize them.... but no one ever reviewed them
fredy has quit [Excess Flood]
<mnemoc> mostly because the polemic regarding how to define "defconfig"
<wingrime> mnemoc: +CONFIG_AXP152=y
fredy has joined #linux-sunxi
<wingrime> mnemoc: thats only must be different
<mnemoc> imo 3.4's defconfig should be "real-life-usage-ready", and obviusly including all sunxi stuff at least as m (for build testing)
<wingrime> mnemoc: agree
<wingrime> mnemoc: but 99% I think a12 config can be droped
<mnemoc> problem is where to draw the line between normal real-life and a bloated kernel
<wingrime> mnemoc: modules
<wingrime> mnemoc: but I not sure how kernel behaves when not found axp
<mnemoc> currently it's handled by script.bin
<mnemoc> if script.bin lies, the kernel dies
<wingrime> mnemoc: what a hell
<arokux2> mnemoc, it is find to build stuff as modules, it won't bloat kernel image.
<arokux2> fine*
<mnemoc> patches welcomed ;-)
<mnemoc> but acks required
<wingrime> already have pending patches)))
<arokux2> mnemoc, I'm busy with AXP mainlining - started already.
<mnemoc> I totally agree is "real-life-ready-defconfigs" with lots of modules
<wingrime> mnemoc: indeed
<mnemoc> wingrime: still 3586 sunxi mails to read
akaizen has joined #linux-sunxi
<mnemoc> wingrime: if you can point me to the important ones, I would appreciate it
<wingrime> mnemoc: but I also, want kernel builds normaly witout any config
<mnemoc> (acked and not nacked)
<wingrime> mnemoc: only bug with sound, but no one commeted it
<akaizen> is there a A20 mkbootfs and mkbootimg?
<mnemoc> good sunxi defaults in Kconfig files are important, yes.
<akaizen> there doesnt seem to be anything realated in sunxi tools
<arokux2> akaizen, what are those?!
<mnemoc> android stuff i think
<mnemoc> and as such, not sunxi-specific
<akaizen> arokux2: I think you're trolling me but its uses to pack the raw boot fs for android (kernel + initramfs)
<arokux2> akaizen, i'm not, never heard of this stuff.
<akaizen> ah ok, I guess my toolchain must be messed up
<mnemoc> akaizen: most people here doesn't use android on their sunxi devices
<akaizen> mnemoc: I'm eventually going to embedded linux but I need a starting point
<akaizen> And current 4K decoding works with this version of android :)
rz2k has joined #linux-sunxi
<akaizen> Android -> Boot2Gecko -> eLinux is the plan
<wingrime> akaizen: 4K on a10?
<akaizen> wingrime: A20
<wingrime> akaizen: on android and stock kernel?
<akaizen> yes
<akaizen> doh
<mnemoc> wingrime: done
<mnemoc> akaizen: mainline linux doesn't support any storage device yet, neither the PMU, or display, or ....
<wingrime> mnemoc: also, all with 3.0 tag
<mnemoc> wingrime: I would prefer to review the issues first instead of blindly closing them all :p
<mnemoc> also, need to go to sleep :(
<wingrime> mnemoc: yes, I look on it
<mnemoc> already got the ftp account for backups. will setup that tomorrow.
<arokux2> mnemoc, +1
<akaizen> mnemoc: thats why I'm waiting to go to linux :) I dont have resources to assist building that stuff out right now
<wingrime> mnemoc: thats can be closed
<akaizen> So anyone know how to go to recovery mode?
<mnemoc> arokux2: btw, that Fan Rong guy posting smp patches for a20 seems to be @allwinnertech
<mnemoc> akaizen: just use our 3.4 for now
<arokux2> mnemoc, why do you think so? he owns cubieboard.
<mnemoc> i think I saw his name in the -dev branches
<akaizen> mnemoc: I would love to but I dont think I can use it for a production product... my boss would kill me
<arokux2> mnemoc, give me a link :p
<mnemoc> and he won't kill you for using 3.11 which mostly only support the very core of the A20 soc?
<mnemoc> arokux2: sleeping time =)
<akaizen> He wont kill me for using the Android that shipped with it and we have support for
<akaizen> Where everything works :)
<mnemoc> akaizen: and is gpl violating, making your product illegal to sell ;-)
<mnemoc> wingrime: need to sleep now :( .... please paste them on /q
<wingrime> mnemoc: ok, tomorrow
<akaizen> mnemoc: thats for the lawyers to decide ;)
<mnemoc> good night
<akaizen> mnemoc: As far as I know, Android allows for proprietary binaries without violating licneses
<mnemoc> akaizen: i'm talking about the kernel and u-boot in there
<mnemoc> akaizen: and cedarx library, which is ffmpeg based :p
<wingrime> mnemoc: you too
<akaizen> Ah yes :) Well the A20 SDK was released
<akaizen> Not sure about cedarx
<mnemoc> akaizen: SDK is gpl violating ;-)
<akaizen> for kernel and uboot
<mnemoc> incomplete, with several parts bin-only
<mnemoc> "to make manufacturers life easier" (tm)
<akaizen> lol, well then shoot im not a lawyer i just want to make it work
<mnemoc> akaizen: sure, but after that explain your boss why it's better to switch to linux and our repos ;-)
* mnemoc off
<arokux2> akaizen, needs some love ^^
<akaizen> arokux2: how so?
<arokux2> akaizen, the more love we get the more open allwinner will be it is as simple as that.
<arokux2> wingrime, what is this?
<wingrime> arokux2: little overclock
<arokux2> wingrime, and? does it work faster?
<wingrime> arokux2: must
<akaizen> arokux2: I'll take a look when I have some more time... under pressure to ship sooon
<wingrime> arokux2: sun4i can be oc to 1.5
<arokux2> akaizen, thanks
<arokux2> wingrime, I care about *code*
<wingrime> arokux2: me too
<arokux2> wingrime, so start to code instead of playing :P
<wingrime> arokux2: I think about pushing hdmi support to uboot
<arokux2> wingrime, go for it. usb is already there.
<wingrime> arokux2: boot menu will be nice
wingrime has quit [Ping timeout: 268 seconds]
popolon has quit [Quit: Quitte]
rz2k has quit []
wingrime has joined #linux-sunxi