rellla has quit [Remote host closed the connection]
<Tsvetan>
hipboi is there something undocumented around the GMAC on A20 so you hide the patches and delay their submission to linux-sunxi until you start selling your board?
<arokux2>
oliv3r, thanks! that is what I eventually have now too
<wingrime>
mnemoc: it looks like a13 drop in replacment
<wingrime>
mnemoc: a10->a20 a13->a23
<shineworld>
always 55nm ?
<arokux2>
mnemoc, there seem to be looots of patches coming :)
<rellla>
wingrime, have you seen benn's mail?
<wingrime>
rellla: I see
<atiti>
anyone has trustzone working?
<atiti>
id like to use it for AES encrypt/decrypt
<wingrime>
atiti: we have no any docs describe trust zone
<wingrime>
atiti: but we have aw security controller
<wingrime>
atiti: that can do
<atiti>
cool, do you have some docs on it? :)
<wingrime>
atiti: see manual
<wingrime>
atiti: a13/a10/a20 have it in user manual
<atiti>
ah
<atiti>
nice
<wingrime>
atiti: but we have no driver for it
<atiti>
are you guys in touch with the allwinner guys?
<atiti>
im asking because I'm in the process of developing an open source product based on the a10, and got in touch with them to get access to some resources
hipboi has joined #linux-sunxi
<mnemoc>
is people happy with sunxi-3.4? can I tag, jump to the latest 3.4.y and spawn a new stage for the rest of the pending patches. or there is something critical pending?
<wingrime>
mnemoc: nothing critical, anything can be done after jump
<wingrime>
mnemoc: but configs need fixes
<wingrime>
mnemoc: sound and otg disabled by default
<oliv3r>
atiti: better + more docs is what we need
<wingrime>
atiti: we still not have a31 manual
<atiti>
oliv3r: but did you guys get the ones allwinner provides?
<oliv3r>
wingrime: the manual does exist, but not officially released :)
<arokux2>
wingrime, otg disabled?! for sun4i?
<oliv3r>
atiti: for a10, a20 and a13 we have them yes
<wingrime>
arokux2: yeax,
<oliv3r>
atiti: but there are missing bits that we would love to have
<arokux2>
atiti, what kind of product it is? very curious....
<wingrime>
oliv3r: can you explian this crap in cedar driver
<oliv3r>
probably not :p
<oliv3r>
but lets see
<oliv3r>
diff -u! :(
<atiti>
oliv3r: alright, just ordered an a10 development kit + all sdks/documentation/source codes, so will see what I can do :)
<mripard>
mturquette: no, you don't :)
<oliv3r>
wingrime: revision A revision B and revision C
<atiti>
arokux2: it will be a wireless screen sharing solution, kinda like what miracast / airplay does, but hopefully a more open and cross-platform version
<wingrime>
oliv3r: thats I see
<wingrime>
oliv3r: but
<mripard>
mturquette: I was just wanting to say to Turl that sometimes, when you don't really have an idea of what you should do, send a hackish patch and see what comes out of the discussion is a very valid strategy :)
<wingrime>
oliv3r: how aw easy renamed a10B to a13b
<wingrime>
oliv3r: it realy revisions or versions?
<mripard>
arokux2: what debug spew ?
<mripard>
Turl: yeah, I saw it as well
jemk has joined #linux-sunxi
<arokux2>
mripard, are those patches usable? :)
<oliv3r>
wingrime: there's a B revision to A10 and a B revision to A13?
<mripard>
Turl: too bad thee's still a long way before getting a proper patch, and that I already started working on it :(
<arokux2>
mripard, "debug spew": sorry, I realize they are there if I enable GPIO debugging, now it is turned off
<arokux2>
mripard, why long way? lots to do?
<Turl>
mripard: there's a new version with the important code included now btw
<wingrime>
oliv3r: I not understang
<wingrime>
oliv3r: I asking you about this stranges in driver
<Turl>
mripard: did I miss a conversation between you and mturquette? :)
<mripard>
arokux2: yes, on deferral, you should cleanup your allocated resources. It's an error after all.
<arokux2>
mripard, right, but you didn't or?
orsic has quit [Read error: Connection reset by peer]
shineworld has quit [Quit: Leaving]
<mripard>
arokux2: what's your bit problem again ? at which address would you have to poke ?
<mnemoc>
wingrime: can you give me a patch to fix the otg/sound issue in all our defconfigs?
<arokux2>
mripard, there are 4 byte. and different bits of should should one poke to enable usb phy. I do not know how these 4 bytes can be shared between two controllers.
<arokux2>
mnemoc, otg is actually a blocker for arm-netbook ppl to use sunxi-3.4, so fix very needed, wingrime !
<oliv3r>
wingrime: well they use the SRAM bits to see what chip they have, they don't use the Version info of the Video Encoder at all
<oliv3r>
wingrime: so could be, that A10B and A10C hav ethe same cedarx etc
<oliv3r>
but fro your diff i find it hard to read ;)
<mnemoc>
only the config thing, nothing polemic please
<mripard>
phew, backlogging done done
<wingrime>
mnemoc: for otg patch are not difficult
<mripard>
Turl: mturquette was asking if he had the reputation of a yeller :)
<Turl>
mripard: ah :)
<mripard>
arokux2: it's very hard to review in its current form, so I don't know.
<wingrime>
mnemoc: but for sound, I not know witch codec ic we need
<mripard>
I'd expect it to
<wingrime>
mnemoc: I simply enabled all)
<arokux2>
mripard, to review what?
<Turl>
mripard: arokux2 spotted a bug on one of your drivers
<mripard>
and yes, I already started working on the SMP part.
<mripard>
arokux2: about the 4 bytes stuff, what are the 4 bytes about?
<mripard>
enabling a GPIO?
<mripard>
a regulator?
<mripard>
just a shared USB register?
<arokux2>
mripard, yes, a shared USB register.
<arokux2>
devm_get_resource will reserve it, so I cannot have another devm_get_resource
<arokux2>
and I cannot claim several bits too :)
<mripard>
arokux2: what register?
<arokux2>
mripard, USBC0_BASE + 0x404
<mripard>
Hmmm, it's not even in the A20 datasheet
<arokux2>
mripard, of course, what you expected? :)
<arokux2>
mripard, but should be poked, otherwise wont work.
<arokux2>
mripard, my ultimate datasheet is the chinese code.
<mripard>
anyway, send your code, we'll deal with that in the review
<arokux2>
mripard, do not want to show the allwinner code to you. it is not with what you want to start your day..... but basically at the offset of 0x404 from the base of USBC0 should be poked by USBC1 and USBC2
<mripard>
but you actually can call platform_get_resource several times
<arokux2>
mripard, good!
<mripard>
it's the request part just after that you can't call several times
<arokux2>
mripard, this one yes! but devm_ioremap_resource only once
<arokux2>
mripard, so now I just ioremap -- can be called multiple times but needs iounmap on cleanup.
<mnemoc>
wingrime: what about changing the defaults in Kconfig instead?
<wingrime>
mnemoc: better talk with techn
<mripard>
arokux2: indeed
<arokux2>
mripard, the other question is the following. the poke at this register should be done based on the controller number (could be 1 or 2). can I have "if" based on address to see if it is 1 or 2? Turl said i'd be an ugly hack and that I need a binding in dev tree
<mripard>
arokux2: and don't worry about my eyes, I already am working on allwinner code this morning
ZetaNeta has quit [Ping timeout: 264 seconds]
<oliv3r>
mripard: RiP
arokux2 has quit [Remote host closed the connection]
<mripard>
that, or you can use the alias in the dt to retrieve the node id
<mripard>
look at of_get_alias_id
<mripard>
(I'm ssh'ing from a train, so excuse the latency :))
<mripard>
of_alias_get_id, sorry
<wingrime>
techn_: ping
<Turl>
mripard: A20 has 5 I2C?
<oliv3r>
Turl: yep
<oliv3r>
Turl: twi0 - twi5
Seppoz has joined #linux-sunxi
<oliv3r>
the former GPS has now twi4 muxed there
<mnemoc>
that's 6....
<oliv3r>
twi4*
<oliv3r>
Phat Phingers
<Turl>
nice
<oliv3r>
a31 has a special hispeed one even
<oliv3r>
so i expect a40 to be have all that too
<mnemoc>
wurstfinger :p
<mripard>
mnemoc: you're such a german now :)
<mnemoc>
:D
<n01>
mnemoc: it is not a compliment
<mnemoc>
I already had ONE! lesson of my german course :p
<mnemoc>
n01: I know, but he knows I'm joking
<mnemoc>
i hope
<oliv3r>
Mmmm wurst, gib mir mahl eine gute sose!
<oliv3r>
or something :D
<mnemoc>
:p
<oliv3r>
wingrime: btw, did you read that we have a new cedarX blob?
<mripard>
mnemoc: good locuk with that
<mripard>
I took german courses back in high school
<oliv3r>
i speak german (austrian really) fluently, i feel shame
popolon has joined #linux-sunxi
* mnemoc
finds french much harder than german... because of the concatenation on the spoken language... and the freaking speed of natives talking
<mripard>
mripard: I'm not even sure Ive been able to understtand a sentence of it at one point
vinifr has joined #linux-sunxi
<mripard>
mnemoc: funny, I find that the spanish actually speak way faster than we do :)
<mripard>
but maybe it's just the language, I don't know :)
<mnemoc>
:)
<mripard>
but yeah, french is pretty hard
<mripard>
mostly because of all the exceptions everywhere
<mnemoc>
that's why we all simply learn english :p
<Turl>
mripard: luckily it's not italian, I heard it's even worse to learn :p
<wens>
I read an article arguing english is hard and broken
<mripard>
ok
<Turl>
yay, cubietrucks and cubie tshirts :)
<mripard>
l'ets just say every language we don't know is hard :)
<oliv3r>
mandarin.
<Turl>
oliv3r: even google has troubles learning that one :)
<oliv3r>
all western languages should be doable; speaking/understanding after some practise, mandarin or cantonese, ouch
<mnemoc>
Turl: italian is pretty easy
<mnemoc>
(for spanish speaking people)
<mripard>
mnemoc: yes, it's french with "i" at the end of each word
<mripard>
(while spanish is french with "o" appended) :))
<mnemoc>
mripard: :p
<Turl>
mnemoc: hm
<wens>
mripard: all latin-derived languages?
<mnemoc>
i heared someone telling "fermatta la ventanatta"
<Turl>
mnemoc: it was brought to my attention that spanish is hard for english speakers because we embed the object pronoun and gender on words
<mnemoc>
thining italian was french + suffix
<mnemoc>
brits and their gender.... problem
<mnemoc>
or it was fermatta la fenetra?
<Turl>
what's fermatta? :P
<mripard>
wens: indeed
<mripard>
Turl: fermer in french is "to close"
<n01>
mnemoc: uat???
<n01>
lol
<mnemoc>
Turl: french for close sounding like italian's halt/stop
<mripard>
I'me pretty sure it was close the window
<mnemoc>
sure
<mnemoc>
but funny
<n01>
chiudi la finestra
<mripard>
see, we speak the same spanish !
<Turl>
mnemoc: I didn't get either the 'close' idea nor the 'window' one :P
<Turl>
'cerrá la ventana' for comparison
lkcl_ has quit [Ping timeout: 268 seconds]
ZetaNeta has joined #linux-sunxi
<mnemoc>
cerrá is only in .ar .... the rest of spanish speaking people tells cierra
<n01>
italian is much more similar to spanish than french
<mnemoc>
ack
<mnemoc>
n01: specially napolitan variant :p
<n01>
hahah yeah, unfortunately I hate napolitans .... I hope the other italians here don't get offended
<mnemoc>
:)
<mripard>
n01: where are you from ?
<n01>
mripard: hometown LECCE but I'm in milan now
<Turl>
mripard: cini sent a new patch, you replied to the old one :P
<wingrime>
jemk: driver change not helped at all
<wingrime>
jemk: there is some problem with TIRG/IRQ/ioctl(wait)
<mripard>
Turl: apart form the file missing, it pretty much still applies though.
<Turl>
mripard: indeed
<Turl>
mripard: I found the virtual/physical timer comments interesting
<Turl>
mripard: looks like we need to add some extra bits on uboot
<mripard>
ok
tzafrir has joined #linux-sunxi
<Turl>
bbl
<rellla>
jemk: ok. summary for a noob: install libvdpau1 libx11 and build libvdpau_sunxi.so should be ok?
<arokux>
Turl, USBCLK and EHCI/OHCI are identical on A10/A20
<mnemoc>
jemk: would you move your project to linux-sunxi's github?
<arokux>
Turl, likely on A31 too.
<mnemoc>
jemk: you are already member of the cedarx team there
<jemk>
rellla: yes, vdpau need X, it doesn't work on framebuffer. but that should be enough
<rellla>
jemk: X server is responsible for output. so i have to get it run!?
* rellla
wonders if xbmc could be take use of it, too...
<jemk>
mnemoc: i would like to wait till it gets a little bit more stable before making it that official
shineworld has quit [Ping timeout: 246 seconds]
<jemk>
rellla: yes, thats limitation of vdpau, it needs X even if it doesn't really use it for anything else then getting the position where to display
<mnemoc>
jemk: less stable is better than nothing....
<mnemoc>
jemk: and you already won the officiality =)
<arokux>
jemk, publishing could spread the world so that you find more contributors / users to test
wingrime has quit [Ping timeout: 245 seconds]
<jemk>
wingrime: could you try measure the time between trigger and interrupt to see if it isn't a software problem
<arokux>
mripard, why don't you CC linux-sunxi when sending patches to kernel-arm? :)
<Tsvetan>
emac is working, for gmac Im not sure as we didnt test yet
<arokux>
Tsvetan, ah, ok. I see now.
\\Mr_C\\ has joined #linux-sunxi
<Tsvetan>
but if it works on cubieboard should be no problem
<mnemoc>
i think the CT will be the first A20 with gmii
<Tsvetan>
yes, it is
<Tsvetan>
A20-SOM-EVB also have Gigabit PHY
<mnemoc>
:o
<Tsvetan>
but the PHY and the camera are not tested yet
<mnemoc>
Tsvetan: and 2GB of RAM? :)
<oliv3r>
Tsvetan: there was a message posted from benn about gmac on A20 cubietruck
<Tsvetan>
too many things to do :)
<mnemoc>
:)
<oliv3r>
Tsvetan: busy man :)
jemk has quit [Ping timeout: 248 seconds]
<Tsvetan>
we always have at least 7-8 things moving in parallel :)
<arokux>
how would I reply to an older e-mail on a mailing list if I wasn't subscribed? i.e. how to reply to a proper e-mail/thread?
<mnemoc>
arokux: gmane
* mnemoc
misses NNTP days
<arokux>
mnemoc, Post action?
<mnemoc>
yes
<arokux>
mnemoc, sure it will send an e-mail to the right thread?
hipboi_ has quit [Quit: Leaving]
<mnemoc>
arokux: if you reply the right mail, yes
<arokux>
oliv3r, Tsvetan saw it and already asked angrily hipboi about :)
<arokux>
mnemoc, I'm confused, because it asks me for a subject.. subject shouldn't be asked for.
<mnemoc>
uhm.... then that'S the wrong option
<oliv3r>
mnemoc: connect to gmane via nntp :p
<oliv3r>
mnemoc: i know i do
<mnemoc>
followup
<mnemoc>
arokux: ---^ that is reply
<mnemoc>
not post
<mnemoc>
oliv3r: :)
<arokux>
mnemoc, yes! thanks!
\\Mr_C\\ has quit []
tinti has joined #linux-sunxi
<Tsvetan>
arokux promoting board as ¨open source friendly¨ and hiding info are mutually exclusively for me, but if they decided to go to this path its up to them, Im sure we will make the gmac working anyway should not be big deal
<Tsvetan>
but shows their attitude :)
<arokux>
Tsvetan, Benn said he'll post patches once the board is out. isn't it good enough?
<specing>
dont go all broadcom on us!
<arokux>
specing, what?
<oliv3r>
Tsvetan: they may have gotten data from AW inside; i'm sure they'll release it, it's just a weird way of doing it; chinese, go figure
Olaff has quit [Ping timeout: 240 seconds]
<Tsvetan>
BTW did someone check RTL8188 doesnt work with current RTL8188
<oliv3r>
not I sorry
<Tsvetan>
Dimitar compiled RTL drivers in module and we try now but it work weird
<specing>
arokux: raspberry pi, the opensource-friendly device, NOT
<arokux>
I think you cannot (and should not) demand from them anything until the board is out.
<Tsvetan>
on some routers it connects without problem
<oliv3r>
the RTL supplied drivers, or the mainline drivers?
Olaff has joined #linux-sunxi
<Tsvetan>
on others it gives shake negotiation errors then connect and works after a while
<Tsvetan>
RTL supplied I think
<oliv3r>
oh, ouch :) good luck :p
<Tsvetan>
oliv3r this sounds evil :)
<oliv3r>
haha; yeah :)
<oliv3r>
vendor supplied uglyness :)
<oliv3r>
THAT is evil :)
<arokux>
why don't they maintain the mainline? I just fail to understand....
<oliv3r>
'NIH'
<oliv3r>
mianline is 'hard' people don't like what we do etc etc
<arokux>
oliv3r, 'NIH'?
deasy has joined #linux-sunxi
<oliv3r>
Not Invented Here
<mnemoc>
AW is product driven, and they only care about their current SoC model. no interest in "long term" support, so they just hack in something to get their latest soc working on the android kernel in use that day
<mnemoc>
usually destroying everything else
<arokux>
mnemoc, what about RTL?
<Tsvetan>
mnemoc with peak lifecycle of 6 to 12 months why should they bother for long term support :)
<Tsvetan>
just waste of manpower
<arokux>
Tsvetan, hm.. SoCs are compatible and once mainlined people send patches and fix their code
pacopad has joined #linux-sunxi
<oliv3r>
Tsvetan: well you plan to support the socs for 5 years don't you
<arokux>
Tsvetan, and why nvidia and samsung do not consider do mainline...
jemk has joined #linux-sunxi
shineworld has joined #linux-sunxi
shineworld has quit [Remote host closed the connection]
<Tsvetan>
oliv3r every product have peak sales where 90% of all SoC sales are made
massi_ has joined #linux-sunxi
<Tsvetan>
the rest 10% can be sold in next 3-5 years but at this poing AW probably will be not interested in any new development
<arokux>
Tsvetan, is it different for tegra and exynos?
Amnience has quit [Quit: Leaving]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<mnemoc>
i think the biggest difference is that the usual AW-based manufacturer is not capable of designing their own PCB
Tsvetan has quit [Ping timeout: 264 seconds]
<mnemoc>
so they provide a reference instead of real docu
<mnemoc>
reference design*
<oliv3r>
bye tsvetan! :(
shineworld has joined #linux-sunxi
<oliv3r>
anyway, i was going to say, while of course that is true, as a company, for your future existance, it's important that your comstumers know you remain dedicated to support your product. while most tablet manufactures probably don't give a shit
<arokux>
mnemoc, meaning nobody is interested in the detailed specs of the SoCs?
<oliv3r>
i guess in the case of AW its moot
<mnemoc>
arokux: only some exotic small buyers like olimex
<mnemoc>
and supporting exotic cases is expensive
<mnemoc>
i assume rockchip is the same
<arokux>
mnemoc, there was guy from Canada who designed a SoM he said he was working closely with AW ppl
<mnemoc>
BDD
<mnemoc>
i have a so-dimm from them
<arokux>
mnemoc, yes. the guy said it should be out very soon :)
<mnemoc>
yeah, he said that 6M ago
<mnemoc>
and bdd seems to be vanished
<arokux>
mnemoc, yes, I've seen it on your cirlces.
shineworld has quit [Remote host closed the connection]
<arokux>
mnemoc, he appeared here several weeks ago and said he left the company but the SoM should appear soon
<mnemoc>
I love the so-dimm som concept... but I can't design carrier boards
* mnemoc
needs to learn PCB design
<mnemoc>
nice
<oliv3r>
lol
<arokux>
mnemoc, so a close work with AW ppl probably substitutes their manual
<oliv3r>
not enough time in your lifetime left
<oliv3r>
i made a few very small ones
<arokux>
I do not like SoMs very much it will make your board huge. why don't sell a design only.
rz2k has quit [Read error: Connection reset by peer]
rz2k has joined #linux-sunxi
<mnemoc>
the SoM includes the expensive part
<jemk>
mnemoc: i moved it, but now i don't have access anymore... looks like you have to add it to the team
<mnemoc>
the carrier board is much cheaper and can be 2/4 layers
<mnemoc>
only
geecko has joined #linux-sunxi
<mnemoc>
jemk: 1m
<arokux>
mnemoc, how many layers is SoM?
<mnemoc>
6
<arokux>
mnemoc, hm, so it means if you had a pcb design of the SoM part copy pasting it to you custom board would make the latter more expensive to produce, because everything should be done on 6-layers PCB now?
Tsvetan has joined #linux-sunxi
<mnemoc>
yes, when you use a SoM it is to make a CHEAP carrier board in smll quantities. when you make a full board is for large quantities
Soru has quit [Read error: Connection reset by peer]
Tsvetan has quit [Read error: Connection reset by peer]
soul has joined #linux-sunxi
soul is now known as Guest84749
Guest84749 has quit [Read error: Connection reset by peer]
atsampso1 has joined #linux-sunxi
atsampson has quit [Ping timeout: 268 seconds]
geecko has quit [Ping timeout: 260 seconds]
<arokux>
A20-SOM will be 15 EUR / 1000pcs, but A13-SOM - 15 EUR / 1000pcs
<arokux>
why so big difference?
<oliv3r>
A13 is more expensive now
<oliv3r>
A20 is i think only 7E for just the chip
<arokux>
oliv3r, your statement contradicts to the price tags
<arokux>
oliv3r, A13-SoM is *cheaper* then A20 one
<rm>
what
<rm>
so both are 15 EUR, where's the "big difference"?
<mnemoc>
arokux: you said BOTH are 15E
<arokux>
oh! A20 is 35, sorry!
geecko has joined #linux-sunxi
<arokux>
A20 - 35EUR, A13 - 15 EUR
<mnemoc>
arokux: because the A20 and it's PMU are more expensive, also more ram, also ethernet phy, ....
<mnemoc>
gmii actually
<arokux>
mnemoc, so gmii is external to the SoC?
<oliv3r>
arokux: no
<arokux>
ok, so it boils down to: memory and A20 vs A13
<arokux>
Allwinner is currently offering its dual-core chip at a price of US$5, an upgraded dual-core model, priced at below US$10,
<oliv3r>
arokux: well mii is an 'interconnect'
<oliv3r>
like a protocol
<mnemoc>
the mac controller is inside the A10/A20 SoC. but the PHY are external
<oliv3r>
so the emac has MII pins, and a PHY has MII pins, and they 'talk' to eachother in mii speak
<ssvb>
arokux: simple demos which use the hardware to show something like spinning cubes, or a hacked variant of quake3 which uses a custom api to drive the hardware and do the same thing as the original version using standard gles
<ssvb>
arokux: the next step is the driver that supports the standard gles api (something that the applications want), which is currently underway - http://libv.livejournal.com/24870.html
<oliv3r>
i think he started hacking on a real mesa driver now
FDCX_ has joined #linux-sunxi
<oliv3r>
a 'drop in replacement' should be available if i'm not mistaekn
<arokux>
ssvb, thanks for explanation. this step of writing driver which supports gles api, is it the last one?
<mnemoc>
it makes no sense to reimplement gl..... a low level mesa driver is enough to get the world rolling
<ssvb>
well, the next step would be fixing bugs and improving performance
<arokux>
mnemoc, what kind of api does low level mesa driver expose?
<mnemoc>
arokux: ask libv :)
<mnemoc>
jemk: sorry for the delay. try now
<arokux>
mnemoc, you seemed to know what you were saying :P
<mnemoc>
arokux: I know what libv has told
<mnemoc>
i'm a parrot
<ssvb>
:)
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<vinifr>
Does someone is writing sunxi TP driver for mainline?
\\Mr_C\\ has joined #linux-sunxi
<arokux>
vinifr, what is TP?
<vinifr>
Touchscreen controller
<arokux>
ssvb, another noob question. the difference of mali to powervr is that nobody has done the hard work for the latter yet?
<oliv3r>
Touch Panel
<oliv3r>
arokux: nobody should want to do that :)
<oliv3r>
arokux: ask libv why :) though it's in the logs :)
<rz2k>
arokux: mali is essentially open (check out mali kernel driver - it gives you everything you need to be able to operate the GPU from userspace - in fact userspace libs do exactly that), powervr has weird kernel driver and lots of layers above in userspace including daemon(!) and other bs.
<oliv3r>
arokux: i think to summarize, it's a piling ball of horse shit
<rz2k>
anyone around with sun7i?
<oliv3r>
rz2k: at home, why?
<rz2k>
oliv3r: sun7i has something really weird happening around 5V for USB and cpu core voltage regulators
<rz2k>
first, they are not getting set
<rz2k>
so you have "probe deferral" errors
<rz2k>
second, if you set them to dummy
<rz2k>
cpu_freq will go insane
<oliv3r>
rz2k: how are you powering your board
<rz2k>
12V power supply
<rz2k>
power supply doesnt matter, this is either .fex in sunxi-boards being wrong or sun7i axp code missing in some places
<arokux>
rz2k, the regulators on sun4i are also deferred, but then turned on eventually
<rz2k>
arokux: today i got a situation when I did not have 5V on USB bus
<rz2k>
and lots of deferrals
<arokux>
rz2k, ok. I get 5V. did you measure or what?
<arokux>
rz2k, normally I see only one deferral
<rz2k>
yes, I both measured and checked with my usb hub and keyboard/mouse
<oliv3r>
rz2k: oh it's stage code; erm well i know a20's powersupply is demanding compared to a10
<rz2k>
I enabled the dummy regulators
<arokux>
rz2k, I see. usb bus gets turned on with a gpio pin
<rz2k>
and now cpufreq doesnt like that because, it cant rule the cpu vdd
<arokux>
rz2k, you can "gpio s PHXX" in u-boot
<arokux>
you are booting sunxi-3.4?
<rz2k>
oliv3r: i emailed my config to Hans, he wrote the regulators code, iirc
<rz2k>
arokux: yes
<arokux>
rz2k, try gpio in u-boot shell and see if the led goes on you usb stick etc. (or measure)
<arokux>
rz2k, (with respect to deferrals: I was talking about mainline kernel)
<mnemoc>
wingrime said techn had the solution, but poor techn is probably still banging his head on the desk after MS's success destroying and buying Nokia for peanuts
<arokux>
why is techn_ poor? has he invested in nokia?
<mnemoc>
saw it on his G+.... I had the same SSD, and also died
<mnemoc>
arokux: works for nokia
<mnemoc>
so he suffered it from inside
<oliv3r>
poor techn_ poor poor techn_
<oliv3r>
his boss is now the new CEO to be of MS :)
<arokux>
linus should really have a raid which dups
<mnemoc>
i hoped Elop returned to MS *before* selling Nokia....
<oliv3r>
bye elop; good luck
<mnemoc>
kudos elop, mission accomplished.
<mnemoc>
desotrying their national pride and giving MS the once best phone manufacturer for peanuts so they can follow Apple's business model
<oliv3r>
heh, so they can try
<mnemoc>
:)
<arokux>
mnemoc, what kind of ssd was it?
<arokux>
mnemoc, Samsung series 840
<mnemoc>
ocz vertex something
<arokux>
?
<arokux>
oh shit
<arokux>
i have the same
<mnemoc>
google a bit. they all died awefully
<mnemoc>
die*
<mnemoc>
in my case the disk controller crashed after transfering 4MB
<arokux>
mnemoc, which one is more probable to live long and happy life?
<mnemoc>
so I connected it to a usb adapter and and an script to loop copying 4MB chunks and pausing for me to reconnect the USB
<mnemoc>
until the 120GB were recovered :p
<mnemoc>
it was a loooooooooong day
<arokux>
mnemoc, are you serious??
<mnemoc>
yes
<mnemoc>
reconnecting the usb worked as "power cycling" the disk
<mnemoc>
reviving it just enough for me to read another 4MB block
<mnemoc>
and so on
<mnemoc>
but recovered my data
<n01>
I have a samsung 840
<mnemoc>
and bought an external RAID1 usb/esata device
<ssvb>
phew, I feel much safer still using the good old mechanical hard drives
<mnemoc>
the main reason is to save laptop battery life
<n01>
and speed
<mnemoc>
but NEVER keep critical data on them
<wens>
the AXP209 page on the wiki needs some cleaning up. the pin description table..... well there is no table.
<mnemoc>
yes, speed is better too. but my thinkpad can last 6-7h which is more important to me than transfer speed
<arokux>
wens, those pages are not actively maintained...
<ssvb>
oliv3r: in nineties, I had a hard drive in my home computer, which was originally a written-off server hard drive from my father's workplace (half of it got covered with bad sectors, probably one of the heads had died)
<oliv3r>
arokux: oh god please forget abou tthat thread, it's 4 months old and chewed through ;)
<ssvb>
oliv3r: but it never died completely in the end :)
wens has joined #linux-sunxi
<oliv3r>
ssvb: lucky you :) mine probably just has some crappy power connection or sata connection
<arokux>
oliv3r, :) already forgotten.
<oliv3r>
those plugs are very reliable
<oliv3r>
not :S
<oliv3r>
arokux: it's a funny read though ;)
<arokux>
oliv3r, yes, Luke is so naive..
<oliv3r>
arokux: luke is here :p
<arokux>
oliv3r, I know
<oliv3r>
arokux: in the end, he was not up to date with the latest developments is all; so it was just some misscommunication
<arokux>
oliv3r, that is what I meant.. trying to convince kernel devs to accept fex? that is so funny to me now ..
<oliv3r>
well i can understand where he's comming from
<oliv3r>
he wanted to please AW into getting their stuff mainlined easier, and they had no plan of letting go of fex really
<oliv3r>
but both parties where miss-informed about how DT works and what it does
<oliv3r>
but since we have most stuff in ML now allready; it's kinda moot now :)
<arokux>
and fex does only a fraction of what the dt deos
<arokux>
does*
<oliv3r>
well if DT hadn't existed, fex was not a bad thing :)
pacopad has quit [Quit: pacopad]
<n01>
uhh is Alejandro here?
<arokux>
oliv3r, so what come out of this meeting if it was at all...?
<mnemoc>
n01: iirc there was something to remove in a Makefile... a gcc arg or something
<mnemoc>
n01: but I gave up on oe long ago
<rz2k>
mnemoc: seems like everyone gives up on oe
<rz2k>
i did that couple days ago
<arokux>
mnemoc, I was going to ask you. how is that project different from a distro like gentoo?
_BJFreeman has joined #linux-sunxi
<n01>
ok :{ I'm fighting with yocto
_BJFreeman is now known as BJfreeman
<n01>
oe is not a distro. you can use it to build one
<mnemoc>
have to run. bbl :(
<arokux>
n01, what kind of flexibility does it add?
<n01>
arokux: is highly modular.
<oliv3r>
hometime
<arokux>
n01, just curious since I didn't get it on their website and you seem to have the expertise. what could be done in oe that cannot be done in highly flexible distros like gentoo and arch linux?
<n01>
arokux: with oe/yocto you define a distro, an image, the packages, architecture, etc ... and what you get is a full distro (kernel+bootloader+rootfs) customized on your board
<n01>
gentoo and archlinux are (sort of) the products
naobsd has joined #linux-sunxi
<arokux>
n01, this can be done with some scripts around gentoo/arch installers. is it those scripts what oe and yocto are all about?
<rz2k>
arokux: oe is like linux from scratch
<wens>
arokux: after cleaning up, my config is still twice as long as multi_v7_defconfig
<n01>
well, yocto/oe use bitbake and "recipes" for each package interpreted by bitbake
<wens>
arokux: it has a bunch of defaults and options selected by other options
<wens>
arokux: correct.
<rz2k>
wens: do you use make savedefconfig ?
<rz2k>
or using your .config ?
<rz2k>
because thats two different things
<arokux>
wens, I cannot understand how you can produce something more then multi_v7. muli_v7 has all kind of SoCs
<wens>
rz2k: I used my .config
<rz2k>
forget about your .config
<rz2k>
use savedefconfig
<wens>
rz2k: so there's a savedefconfig, I was wondering if there was a script or not
<wens>
rz2k: thanks
<rz2k>
it will save your current .config to file named defconfig
<rz2k>
with everything sorted and cleared
<rz2k>
thats why you have *_defconfig
<wens>
much better, 61 lines
<rz2k>
arokux: Arch is a distro, with rules and conventions (used package manager, updates and whatever). OE is a build system, you define everything yourself, including package manager and everything else.
<rz2k>
whats bugging you in OE, it even says open EMBEDDED
<rz2k>
clearly headed for embedded use :p
<arokux>
rz2k, n01 you can even define a package manager? :)
<rz2k>
yes
<rz2k>
opkg, rpm or apt
<rz2k>
rpm by default, in yocto
<n01>
arokux: yep
<n01>
arokux: if you are interested in the topic I highly suggest to take a look to buildroot
<n01>
much more usable than yocto
<rz2k>
if only yocto had better documentation
<arokux>
n01, I've used buildroot and understand why is it out there. it builds you a system with small footprint
<n01>
arokux: yocto/oe is buildroot on steroids
<arokux>
n01, rz2k so yocto is a-la buildroot?
<rz2k>
sort of, but bigger
<rz2k>
and with own ideas
<arokux>
it is funny people want to be able to configure package manager, though
<n01>
like yocto-kernel-style
<arokux>
n01, rz2k thanks guys
<n01>
glad to help
<rah>
oe is a nightmare
<rah>
it's a moving target
<rah>
a continuously moving, continuously breaking target
<n01>
yocto is even worst if possible
<wens>
arokux: just sent my defconfig to the ml
<rah>
err.. yocto *is* oe isn't it? :-)
<n01>
rah: no :) oe is now *compliant* to yocto
<arokux>
wens, thanks, will try it out later tonight
<rah>
"In March 2011, the project aligned itself with OpenEmbedded, an existing framework with similar goals, with the result being The OpenEmbedded-Core Project."
<rah>
oe-core == yocto
<n01>
rah: oe-core != oe
<rah>
n01: oe-core ~= oe
<rah>
:-)
<n01>
"Prior to version 1.0, the Yocto Project has integrated recipes and other build metadata from the OpenEmbedded project through the Poky build system, which has been in partnership with OpenEmbedded for many years"
<n01>
with yocto you can use recipes of oe and oe-meta itself
<n01>
anyway it's fuc**g hell
boycottg00gle has left #linux-sunxi ["ERC Version 5.3 (IRC client for Emacs)"]
<WarheadsSE>
well considering that almost everything else roots back to the AOSP as a base .. I still think I am right
<mripard>
WarheadsSE: yet, the linux foundation consider AOSP has a linux distribution :)
<WarheadsSE>
Well, please point me to who in the world shares the same view of everything?
<mripard>
hence why I ask what is your definition of a distro :)
hipboi has quit [Quit: Leaving]
<WarheadsSE>
a distribution is a userland bundle
<WarheadsSE>
kernl/distro are not linked.
<mripard>
great then, we have the same view on this
<WarheadsSE>
the android kernel is now also a linux kernel, but the userland is quite seperate from the general linux userland
<mripard>
so how is Android not a userland bundle ? :)
<arokux>
so no package management and repos?
<rah>
"general linux userland" I think you mean "GNU"? :-)
<WarheadsSE>
if you aren't actually using package management && a repo, I wouldn't call it a distribution so much as an image based on a choice of packages
<WarheadsSE>
rah, usually it is the gnu toolset&chain
<WarheadsSE>
from there its completely AWOL as to where you take it
<arokux>
on top of which linux will be the yet to be released android-4.4 based?
<WarheadsSE>
<.<
<WarheadsSE>
linux is a kernel
<arokux>
WarheadsSE, yes
<WarheadsSE>
that is all
<WarheadsSE>
anroid 4.4 will probably need some kernel features, but 99% of the change will be their userland
<arokux>
WarheadsSE, google takes some kernel and obfuscates it. so which one will be under 4.4?
<mripard>
arokux: obfuscates? how so?
<libv>
i am probably asking a total newbie/RTFM question here, but has anyone succeeded using the mali on a20 with sunxi stuff?
<WarheadsSE>
arokux: I don't work for google
<arokux>
mripard, that was just a joke, do not take it seriously. but partly because (as far a i know) google is upstreaming very little
<WarheadsSE>
libv: I haven't tried..
<WarheadsSE>
arokux: they tend to upstream *kernel* changes over time
<mripard>
arokux: on a general basis, it's not true at all.
<mripard>
about Android, it used to be
<mripard>
it's not so true anymore either
<mripard>
pretty much everything but ION is now in mainline
<arokux>
mripard, ok, thanks for the update
wingrime has joined #linux-sunxi
<mripard>
and the gap between ION and the rest of the kernel tends to thin a lot
<mripard>
(ie, when it was introduced, it was really different, now it uses a lot of in-mainline APIs, making it quite thin, compared to what it used to be)
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
<mripard>
but the only thing Android is really concerned about is Kernel <-> userspace APIs, and that part is pretty much done. That doesn't mean that the support for your device is in mainline
<arokux>
Android 4.3 is based on 3.4.10
<mripard>
just that the kernel has the features needed by android.
<mripard>
and the kernel usually isn't updated, at least for the nexuses, during their lifecycle
iamfrankenstein has quit [Ping timeout: 268 seconds]
<mripard>
both my nexus 7 and my nexus 4 run android 4.3, yet the nexus 7 uses a 3.0 kernel, while the nexus 4 a 3.4
<mripard>
as long as you have the right features in the kernel, the kernel version doesn't matter at alle.
<mripard>
-e
<arokux>
I'd think asus, samsung and co. should be angry at google for updating there nexuses so often
<arokux>
but they are not and even produce hardware...
<hno>
wingrime, yes? (better to ask immediately)
<wingrime>
hno: 256 Mb dram conf workable?
<wingrime>
hno: a13 som will have only one ddr3 ic
grevaillot has quit [Remote host closed the connection]
<wingrime>
hno: and lowend variant will have only 256Mb
<wingrime>
techn_: ping?
<wingrime>
techn_: ping
ganbold_ has quit [Remote host closed the connection]
<hno>
wingrime, yes. A13 OLinuXino MICRO only have 256 and works. But using MALI on that is a bad idea I guess (not tested).
<mnemoc>
with good defaults we can minimize the defconfigs a lot
<lkcl_>
rah: it's aaaall goood :)
<arokux>
lkcl_, :)
<arokux>
lkcl_, have you seen last pearl from Linus?
<lkcl_>
rah: i ignored the trolls [if that's what they're called]. by the end of that thread i had enough information to get back to allwinner's directors
<lkcl_>
arokux: i haven't.
<lkcl_>
rah: that was the purpose of raising that discussion, basically.
<lkcl_>
arokux: what's been going ooon? tellme, tellme :)
<wingrime>
lkcl_: ?
<hno>
wingrime, I don't have an opinion on u-boot framebuffer support. But don't see the point. Better to get kernel up and running as fast as possible, and if you are messing around then you better have a uart anyway as you will be quite lost without a good input device.
<ssvb>
libv: yes, mali works on cubieboard2
<lkcl_>
wingrime: i'm as lost as you at this point - i'm trying to establish context (just got in)
Tsvetan has joined #linux-sunxi
<jlj>
wingrime: ah okay, thanks
<libv>
ssvb: anything special needed for that, or just the normal stuff?
<ssvb>
libv: it just works out of the box if using the stage/sunxi-3.4 kernel, nothing really different from a10
<libv>
ssvb: cool, seems like i should just kick cwabbott up the rear a bit then :)
<libv>
thanks
<ssvb>
libv: I actually was about to ask a newbie/RTFM question about how to make sure that the second PP is used, because right now the performance results indicate that this is not the case
<ssvb>
libv: I'll post more details to the mailing list
<libv>
ssvb: LD_PRELOAD the wrapper, and send me the result
<rah>
lkcl_: it seemed very much like the purpose of that thread was not to gather information for the purpose of improving linux but instead, to create a situation where others acknowledged what you saw as a privileged position of communicating with the directors of Allwinner, for the purpose of boosting your ego
<lkcl_>
rah: i just had to filter that shit out.
massi_ has quit [Quit: Sto andando via]
<lkcl_>
it was actually that i happened to *have* a privileged position of access to the directors of allwinner, and i was keenly aware that i was completely unprepared for it.
<lkcl_>
so, even knowing that many people regard me as being a complete fucking moron, and that several people would _again_ get completely the wrong idea, i had to risk getting onto several public mailing lists, filter out the shit and get the information that i needed.
<lkcl_>
that was successful, and i've *already* left it behind.
<lkcl_>
rah: if i was after an ego boost, i would have written up a follow-up message saying "hey here's how i'm greater than all of you, fuck you lot, blah blah etc. etc." - wouldn't i?
<lkcl_>
as it was, the responses were _so_ outrageous that i decided *against* making a follow-up report
<lkcl_>
not least because some fucking idiot registered my email address over 1,000 times with a 3rd party service in an attempt to spam-bomb my email address.
<arokux>
lkcl_, did the meeting actually happen? with what result if so?
<wingrime>
arokux: ?
<lkcl_>
arokux: of course! it went very well. it was only with mr ding. we have a lot of follow-up discussions to do, but i was able to get across many of the good points that the _sensible_ people raised.
<arokux>
wingrime, what?
<wingrime>
arokux: where I can read
<lkcl_>
in the end we only had about an hour, hour and a half, so it was difficult to cover everything.
<arokux>
wingrime, read what?
<lkcl_>
but i got across that they should never have used powervr, i mentioned libv's work on lima driver
<arokux>
lkcl_, didn't they new about it?
<lkcl_>
mentioned that they should investigate Tegra licensing for 3D
<lkcl_>
arokux: they chose powervr because of the increased performance: mali's propriertary libraries have a *massive* performance hit.
<wingrime>
arokux: I not understand what are you talking about aw directors
<lkcl_>
and no! mr ding is a hardware engineer, not a software engineer.
<lkcl_>
listen guys i _really_ gotta get back to doing the touchpanel on the flying squirrel
dwilkins has quit [Ping timeout: 264 seconds]
<lkcl_>
arokux feel free to update wingrime on the context :)
<arokux>
wingrime, there was a thread where lkcl_ tried to convince kernel devs to accept fex etc.
<arokux>
wingrime, i mentioned it today..
<arokux>
lkcl_, last question.. have you asked for docs? :p
<rah>
lkcl_: egoism isn't necessarily about being greater than others, so being after an ego boost doesn't imply sending a follow up message stating that you're greater than others; in fact, even if it *was* about being greater than others, the ego boost doesn't depend on telling others about that boost
<lkcl_>
arokux: yes!!
<lkcl_>
i bloody well did
<mnemoc>
wingrime: It reflects the sound problem and also some emac...
<wingrime>
mnemoc: yes, we need more one patch for sound
<mnemoc>
wingrime: pushed the OTG commit
<lkcl_>
rah: i'm sorry, i have to focus. i've already moved well beyond that discussion - it served its purpose, i have to leave it at that, i have too much to get done. sorry
<mnemoc>
i was just default m it
<arokux>
lkcl_, so where are they? all sunxi.org gets is from third parties :p
<mnemoc>
would*
<lkcl_>
does anyone know how to get ft5306 up-and-running
<lkcl_>
arokux: *sigh* i didn't say he actually responded and took _action_....
<wingrime>
lkcl_: it have vendor tool for firmware backe
<lkcl_>
arokux: but... he now knows. i also mentioned the importance of getting info on libcededarx
<lkcl_>
wingrime: ouch. any ideas where that is?
<wingrime>
*bake
<lkcl_>
argh argh out of time - can i ask you a favour, wingrime: could you mail me (to arm-netbooks) anything you know?
<wingrime>
lkcl_: better not use firmare in opensoure tablet
<lkcl_>
i'll be back in about 2 hours
<wingrime>
lkcl_: I will be here
<lkcl_>
wingrime: *sigh* too late - this is a $7 touchpanel, everything else is $11 to $19
<lkcl_>
if i'd known i would have looked harder or pushed it more.
<lkcl_>
blagh.
<lkcl_>
ok. gotta go.
<lkcl_>
thanks wingrime
<hno>
wingrime, yes, with two different generations of drivers.
<wingrime>
arokux: are you played with musb?
<ssvb>
libv: at least limare_cube_companion_spinning with 1280x720-32 monitor resolution is showing ~220 fps for 2 PP and ~145 fps for 1 PP (when the second PP is commented out from the table in the kernel) on cubieboard2
<ssvb>
libv: but egl_cube_companion_spinning is showing the same ~145 fps in both cases :(
<arokux>
wingrime, just to see if it works. jukivili actully works on the code.
<ssvb>
libv: at least the kernel seems to be fine, but the blob is probably not
<Turl>
ssvb: is it using a sun4i blob?
<ssvb>
libv: I'll try to make a log with the wrapper
<ssvb>
Turl: yes, the old r3p0 mali blob
<arokux>
wingrime, why were you asking?
<wingrime>
arokux: add support to uboot
<wingrime>
arokux: if we will have usb+disp
<wingrime>
arokux: boot menu will be awesome
<arokux>
wingrime, you do not need musb to add usb to u-boot
<Turl>
Please contact support-mali@arm.com with the shader causing
<Turl>
the problem, along with this error message.
<arokux>
wingrime, afaik, usb host controllers are enough
<Turl>
the things strings can show you :p
<wingrime>
arokux: expose nand like usb mass device
<arokux>
wingrime, ah, ok. btw, there are directories musb and musb-new in u-boot tree
<wingrime>
arokux: yes, usb keyboard + hdim , and cool boot menu
<arokux>
wingrime, usb host can be added easily to u-boot by me
<arokux>
wingrime, will you add hdmi?
<nove>
wingrime, jemk, NEW at the trace viewer, a more compact representation (left click to open/collapse), plus large traces speedup
<vrga>
'lo folks, i'm apparently a derpmeister and have no idea how to use git. i checked out the linux-bsp stuff and i'd like to update the linux-sunxi thing to latest.
<vrga>
how would i go about doing that?
atiti has quit [Ping timeout: 260 seconds]
<rz2k>
yay
<rz2k>
no one could guess why I have regulators not showing up
<rz2k>
solution is: compile i2c support built-in
<rz2k>
in sun7i_defconfig i2c is 'm'
<rz2k>
wtf? :)
<buZz>
most defconfigs seem morronic
<buZz>
the cubieboard2 one doesnt support swap or partitions by default
<ssvb>
rz2k: the sun7i_defconfig defconfig is also not enabling EMAC
<rz2k>
yes, also that
<rz2k>
we should check all defconfigs before tagging new release
<rz2k>
also good night
rz2k has quit []
<wingrime>
nove: screen shot please
<wingrime>
arokux: I thinking about adding HDMI to uboot
<Turl>
wingrime: if uboot can configure framebuffer you can use "simple fb" driver on linux
<Turl>
at least until a proper one is in place
<wingrime>
Turl: i want add hdmi basic, for better understand disp, before we will do mainline support