ganbold has joined #linux-sunxi
anunnaki has quit [Remote host closed the connection]
anunnaki has joined #linux-sunxi
rellla has joined #linux-sunxi
<mnemoc> rellla: for "replies" prefix the paragraphs with : ... one per level ;-)
<mnemoc> helps to make wiki discussions readable
<rellla> ok. *grmpf* wiki noob
<rellla> i wondered myself, why i couldn't find my statement after reopening ;)
<mnemoc> *g*
<mnemoc> also, --~~~~ translates to a wiki signature, with links and timestamp
<mnemoc> there is a "button" to inject the signature too
<rellla> fine. better now.
<mnemoc> :)
<mnemoc> any suggestion for the /topic ?
<mnemoc> it looks kind of boring
<rellla> something like: Allwinner/sunxi development discussion - See | | Logs at
<rellla> I'm fan of the friendly "- Don't ask to ask. Just ask! -" - slogan
ChanServ changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See | | Logs at
<mnemoc> like that?
<rellla> like that.
<rellla> if you find it to boring, you can alter it with a new saying every day :P
<mnemoc> :)
<Turl> rellla: ouch
<rellla> Turl: think about your answer ;)
<Turl> rellla: I wonder if he really "wrote and researched" those pages
<rellla> maybe he joins here. better discuss it in irc. wiki-talk is even more stressful than email-talk.
<Turl> yeah
<mnemoc> long life to irc
<Turl> If he really did write those pages, we can restore the contents so that they can be contributed to another more suitable wiki, like elinux
<Turl> but they don't really belong to linux-sunxi, at least not without sunxi specific stuff in them
<mnemoc> but we do need some short page about the PVR GPU in A31... linking to external resources for details
<mnemoc> even if it's 2 lines and a link
<Turl> the A31 page? :)
<rellla> mnemoc: #arm-netbook was my first contact with irc. it's kind of VERY cool but extremly addictive \o/
<Turl> link to wikipedia
<Turl> rellla: haha
<mnemoc> the guy that made the A10 sodimm claims to have preferential access to allwiner, linux drivers for proprietray A31 stuff and that CAN, SATA and CEC registers documentation is on it's way
<mnemoc> rellla: been non-stop on irc since 1995 :) it is addictive :p
<Turl> for A10 or A31?
<mnemoc> a31
<mnemoc> back in 20m
<Turl> A31 has sata? :)
* rellla hopes, that mnemoc has seen some stars, clouds and sun since 1995
<Turl> :p
<mnemoc> Turl: two different statements there
<mnemoc> Turl: 1. they claim to have linux drivers for the PVR of the A31
<mnemoc> 2. they said allwinner is going to send them CAN, SATA and CEC documentation for A10
<mnemoc> rellla: I have a window looking to the harbour next to me :)
<Turl> mnemoc: CEC would be nice
<Turl> mnemoc: CAN is like the forever forgotten feature of A10 :p
<Turl> mnemoc: GPS docs would be awesome
<mnemoc> tom said some time ago allwinner itself doesn't have access to the sources of the gps driver, it's controlled by the ip owner
<mnemoc> BUT
<mnemoc> we have a .ko with debugging symbols
<Turl> so they embedded a gps receiver which they don't know how it works
<Turl> awesome!</irony>
<Turl> mnemoc: is that a beach I see on the horizon?
<mnemoc> no, another harbour :p
<mnemoc> just checked my irc logs. if you want to add it to [[A10/GPS]], the owners of the IP are
<mnemoc> and iirc it requires an RF companion chip, like the ethernet IP
<rellla> mnemoc: beautiful! no harbour, no beach, no mountains here. only a few cute facades of historical buildings 20mtrs away. and a church - whoaa ;)
<mnemoc> :)
<rellla> has tom already had his "date" with allwinner?
<mnemoc> no idea
<mnemoc> if you look at the bottom of the picture you'll see how respectful of the transit laws spaniards are
<Turl> parking in the middle of a turn in a highway?
<mnemoc> :)
<Turl> only in spain :p
<mnemoc> some times even in second row
<mnemoc> Turl: also, it's not a simple turn... it's roundabout
<mnemoc> s/in second row/double-parking/
<Turl> we say 'in second row' here too :p
<mnemoc> while googling the term I saw the brits don't even have a term for it
<mnemoc> double-parking is a US term
<Turl> mnemoc: I'm looking at some of the clock gatings, but can't make sense of some initials
<Turl> any idea what MP, DE-FE, DE-BE can mean?
<mnemoc> none at all
<Turl> ACE? MS?
<Turl> A10 has all the things, even PATA apparently
<mnemoc> all winners put together in a cheap chip :)
<Turl> MS might mean memory stick :p
<mnemoc> makes sense
<Turl> heh, it has PS/2
<mnemoc> look at the user manual, they have a huge list of content-less tiyles
<mnemoc> titles*
<Turl> 2 PS2 controller for connecting external PS2 mouse and PS2 keypad
<Turl> it even has a crypto accelerator
anunnaki has quit [Ping timeout: 264 seconds]
anunnaki has joined #linux-sunxi
von_fritz has quit [Ping timeout: 264 seconds]
ganbold_ has joined #linux-sunxi
<libv> hah, i still have debug code in there.
* libv should shore up the libump versioning.
<mnemoc> libv: what about using an script to detect ABI instead of gnu make voodoo?
<mnemoc> a shell case can deal with different -dumpmachine much better
rellla has quit [Remote host closed the connection]
<libv> it's not that much voodoo
<libv> let's first see what Eric his gcc -dumpmachine tells us
<libv> and i am sure that a handful of wildcarded statements will suffice under make
<mnemoc> case "$machine" in *-gnueabihf) ABI=armhf;; ..... esac
<mnemoc> constructs like foo-*-bar|something|*-narf) are trivial on sh, ugly in make, and... gnu extensions :)
<libv> how would one embed this into the makefile?
<mnemoc> ABI=$(shell ./
<libv> because my make-foo is non-existent and immediately forgotten afterwards
<libv> ah, so a separate script indeed
<mnemoc> yes
<mnemoc> it can also do the empty ABI ranting
<mnemoc> echo "...." >&2 ... or even cat <<EOT >&2
<libv> i'd rather keep that unified across all three arguments
* mnemoc is a shell monkey
rellla has joined #linux-sunxi
hramrach has joined #linux-sunxi
techn_ has joined #linux-sunxi
<Turl> wouln't it be sufficient to check for 'hf' in the name?
<Turl> and maybe arm, if you want to avoid people running it on x86?
<Turl> $(findstring hf,$(findstring arm,dumpmachineoutputhere))
<Turl> nevermind that won't work
* mnemoc insists a shell script can do it better
<Turl> libv: mnemoc
<mnemoc> i hate gnu-isms in makefiles with all my guts, so I prefer to not comment
<bfree> now has a kernel with a hugely expanded config ... of course it doesn't do anything "nice" like manage to generate uImage/uInitrd yet, or even mksunxiboot the spl (or have sunxi-tools to provide fexc) ... and all only tested on cubieboard :-/
<mnemoc> bfree: it would be awesome if you could join up efforts with zumbi
<bfree> I agree ;-)
<mnemoc> bfree: sunxi-mali is begging to be packaged
<bfree> so is sunxi-tools which is a lower-hanging fruit ... and can go in main not non-free ;-) ftr afaik in a "real" repo I'd have to put the current kernel in non-free (at least due to some of the wireless stuff crammed in)
<mnemoc> sunxi-tools is totally free
<mnemoc> and most of sunxi-mali as well
<mnemoc> sunxi-mali is -dev (free) and -lib (non-free)
<mnemoc> well... the resulting is free
Turl has quit [Excess Flood]
<mnemoc> sunxi-tools needs restructuring and decent configure/makefile before proper packaging
Turl has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
anunnaki has quit [Ping timeout: 248 seconds]
<mnemoc> guess what found in the GPL violating A10S 1.4 SDK Tsvetan gave for garbage? drivers/video/sun5i/hdmi/aw/hdmi_cec.c
<RaYmAn> oooh
<techn_> ooh!
<techn_> where where?
<Turl> heh nice :)
<RaYmAn> usual high quality allwinner code ;)=
<Turl> yeah, and no license headers :<
<techn_> no copyright :(
<techn_> but binaries are somewhere to re ;)
<mnemoc> there is enough evidence that the lack of proper gpl headers in allwinner's code is just an unintential mistake and can be corrected as bug
<mnemoc> *g*
<Turl> this channel is publicly logged mnemoc :p
<mnemoc> they said that when tom fixed the headers on our tree
<Turl> heh, really?
<mnemoc> allwinner itself
<mnemoc> sorry, the code is GPL, we just missed the formality
<mnemoc> then tom asked for permission to fix the mistake in our tree during his not-work hours, and they agreed
<mnemoc> that's how our tree got formally GPLed
<mnemoc> next to that he was asked by his boss to remove the allwinner word from their internal tree :<
<mnemoc> and Reuuimlla Technology Co. was "created"
<mnemoc> (note it has the same letters as allwinner)
<mnemoc> mostly
<mnemoc> w -> m
<RaYmAn> upside down, lol
<mnemoc> yes
<mnemoc> the A13/A10 1.5 SDKs say Reuuimlla. some months later (before the A10S 1.4 SDK) they switched back to Allwinner
<mnemoc> strange company
<mnemoc> but anyhow, the point is that they know (and agree) all that code is GPL.... and the reason to include some parts as .ko/.o is not to hide it BUT to avoid manufacturers making changes that could affect the reliability of the product...
* mnemoc better shuts up and returns to $work$
<mnemoc> Turl: please invite mripard :)
ZaEarl has joined #linux-sunxi
ZaEarl has quit [Read error: Operation timed out]
<hramrach> mnemoc: you are running GNU make so you can't do pretty much anything without GNUisms
<hramrach> all meke implementations are mutually incompatible
mripard has joined #linux-sunxi
<mnemoc> hramrach: :)
<mnemoc> mripard: welcome!
<mripard> thanks
<mripard> I somehow overlooked at the creation of this chan
<mnemoc> it only became active yesterday
<mripard> ah, that's why
<mnemoc> :)
theRat has joined #linux-sunxi
<Turl> mripard: I added proper output names for all the gates
user135 has joined #linux-sunxi
<Turl> there are a lot of them :p
<mripard> what do we need this for ? (no idea, just asking)
<Turl> mripard: to register the gates we need a unique name
<mripard> hmmm, are you sure
<Turl> yeah, if you repeat the name or use NULL it errors
<Turl> the name is used on the debugfs representation and for clk_get and the like
<mripard> ahh, when you do a clk_get?
<mripard> hmmm, it's weird
<Turl> I'm sure the name is needed because I was getting screenfuls of errors until I sprintf'd unique names for each one
<mripard> mvebu indeed seems to be reading from the dt to register
<mripard> but never declare a clock-output-names property
<Turl> mvebu uses the value from a table and optionally overrides with the dt string (which is not defined)
<Turl> I assume the table is from when mvebu was a non-DT platform
<mripard> mvebu has never been a non dt platform :)
<Turl> hm
<Turl> why would you override your own table with dt then?
<mripard> I have no idea
<mripard> but mvebu has been created to integrate the armada SoCs
<mripard> and merge into it the kirkwood, orion and dove platforms that were not using the dt
<mripard> so mvebu has been from the start using DT
<mripard> the legacy code is in mach-orion, mach-kirkwood or mach-dove iirc.
<mripard> anyway, ok, it seems like the right way to go
<mripard> but, again, you should really send a v2
<mripard> it's been a month now
<Turl> I will
<Turl> when does -rc1 close?
<mripard> it's kind of a long time between two versions
<mripard> around the 4/3
<mripard> but you can send it before
<Turl> that's a bit too far, yes
<Turl> I'm working on top of torvalds/master now that the patches that killed the old driver were merged
<Turl> because we need to readd bits from there
<mripard> just apply this patch on top on whatever 3.8 tag
<mripard> and rebase your work on top of this
<mripard> you don't have to use master
<mripard> which should be pretty unusable right now anyway :)
<Turl> worked fine when I tried earlier today
<mripard> you were lucky then :)
<Turl> I'll send the v2 on monday morning after I get a chance to test on the board and make sure I didn't break anything
<Turl> need to update the commit message too
ZaEarl has joined #linux-sunxi
<mripard> ok, great
<Turl> I pushed sunxi-clk-7 tag on my git repo if you want to have a look later
<Turl> it has the gates code on separate patches
<mripard> what's your git URL?
<Turl> I did a little bit of cleaning and refactoring on the C part, so it's only compile tested
<mripard> you will only send commits 887cee8 to 42365fd, right?
<Turl> mripard: depending on if it works, I might send the other two on the series, but with RFC tags and cc Gregory and Mike on them
<user135> hi, I've just compiled u-boot-sunxi for sun5i, and attempted to boot CWM, but the screen is blank
<user135> I'm not sure if I compiled u-boot-sunxi ok
<user135> I used the embeddian toolchain
<Turl> compiling uboot is pretty trivial; make device_config; make
<user135> oh, that's probably why, I forgot make device_config :)
<user135> thanks
<Turl> (device is your device name, not the literal word)