hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<leviathanch_> Turl: I will isolate the exact patch
<leviathanch_> and return with a solution
<Turl> leviathanch_: might be one of the bugs I fixed
<Turl> leviathanch_: when you prepare & enable which clock?
<leviathanch_> it's not even only then
<leviathanch_> the problem persists when I fully disable the mmc host driver in menuconfig
<leviathanch_> I checked the registers
<leviathanch_> the clocks have the same registers in A20
<leviathanch_> so I can't be that
<leviathanch_> *it
<leviathanch_> or maybe I just found out what "not yet tested on <chip>" can mean in the worst case
<leviathanch_> >_>
<Turl> leviathanch_: try adding 'clk_ignore_unused' to the cmdline
<leviathanch_> ok
<leviathanch_> the following: as long as the device tree of the A20 doesn't reference it
<leviathanch_> nothing happens
<leviathanch_> (of course)
<Turl> doesn't reference what?
<leviathanch_> the clock drivers
<Turl> (also, make sure you got this patch, otherwise the option doesn't do anything
<leviathanch_> pll6 is a dummy clock for a20 dtsi
<leviathanch_> in sunxi-next
<leviathanch_> and you haven't changed that yet
<Turl> I have in my current tree
<leviathanch_> oh
<leviathanch_> ok
<leviathanch_> that might explain a lot
<leviathanch_> it started booting after I had replaced the dtb by the one from sunxi-next
<leviathanch_> which doesn't enable the pll driver
<leviathanch_> Turl: I have to sleep now
<leviathanch_> already 2.a.m.
<leviathanch_> it would be great if you could cross check if my suspicion is correct and there is some really neasty thing going on with reinits of plls on A20
<leviathanch_> just code side
<leviathanch_> I can test it
<leviathanch_> ;-)
<leviathanch_> n8
<Turl> night
uwe_ has joined #linux-sunxi
robb83 has joined #linux-sunxi
robb83 has quit [Client Quit]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<robertjw> I've installed sunxi/fedora19 on a mele a2000, and now I'm looking for the best way to bring up xbmc on it. The fedora repo does not appear to have the package.
hipboi has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
bamvor has quit [Remote host closed the connection]
libv has quit [Ping timeout: 240 seconds]
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
libv has joined #linux-sunxi
wingrime has joined #linux-sunxi
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
wingrime has quit [Ping timeout: 246 seconds]
JohnDoe_71Rus has joined #linux-sunxi
robb83 has joined #linux-sunxi
ykchavan has joined #linux-sunxi
akaizen has joined #linux-sunxi
robb83 has quit [Quit: - A hand crafted IRC client]
Soru has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
<mnemoc> arokux1: done
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
atiti has quit [Ping timeout: 276 seconds]
geecko has joined #linux-sunxi
BJfreeman has quit [Quit: had a good time]
<oliv3r> lo
cajg_ has joined #linux-sunxi
cajg has quit [Ping timeout: 256 seconds]
cajg_ is now known as cajg
n01 has joined #linux-sunxi
shineworld has joined #linux-sunxi
massi has joined #linux-sunxi
<geecko> hi
<shineworld> hi geecko
panda84kde has joined #linux-sunxi
<JohnDoe_71Rus> low
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier.]
JohnDoe_71Rus has joined #linux-sunxi
focus has quit [Remote host closed the connection]
VicenteH has quit [Ping timeout: 264 seconds]
wingrime has joined #linux-sunxi
arokux1 has quit [Quit: Leaving]
arokux has joined #linux-sunxi
popolon has joined #linux-sunxi
<arokux> hi all
<oliv3r> mornin'
<oliv3r> almost done with my bookkeeping from 2011-2013; then it's back to hacking!
<arokux> :)
<arokux> I wonder how should I debug my usb for mainline further
tzafrir has joined #linux-sunxi
<tzafrir> I'm looking at buying a cupieboard2 or a mele unit. I just wonder what exactly the packaging contains.
<tzafrir> does not provide much information
<tzafrir> Specifically, do those contain an AC/DV power adapter? usb2serial cable?
<arokux> tzafrir, which country are you at?
<oliv3r> tzafrir: for hacking, a olimexino is very good
<tzafrir> arokux, Israel
<oliv3r> tzafrir: if you want to use it 'also' as a set-top-box etc, the mele is reasonably hacking friendly
<oliv3r> arokux: matson is matson hall, from cubietech
<oliv3r> former AW employee
<arokux> tzafrir, too far :( I maybe considering to sell my Mele A1000 (A10 based)
VicenteH has joined #linux-sunxi
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
<wingrime> ssvb: ping
<rm> tzafrir, check if it's not cheaper to buy at for you
<rm> there it's usually quite clear what the deal includes
<tzafrir> I could not find a way to keep my purchace non-public in aliexpress. Which is why I avoid it so far
rz2k has joined #linux-sunxi
<JohnDoe_71Rus> what kind RTC module will be beter for cubieboard?
<rm> tzafrir, it shows e.g. me as R***n M. in the feedback section
<rm> nothing else is public...
<arokux> is it really safe to buy something like cubieboard on aliexpress?
<tzafrir> It shows me as Alibaba-express-is-bad-for-privacy in the the name. I thought this would help me wit htheir support.
<tzafrir> No such luck.
<tzafrir> I still would not buy from them just because of that
<tzafrir> arokux, from what I understand it, the shop is operated by the same guy
<tzafrir> I asked here (or rather: on #arm-netbook) before buying there...
<JohnDoe_71Rus> tzafrir: you can by in store witch accept paypal
arokux has quit [Remote host closed the connection]
arokux has joined #linux-sunxi
<hramrach> well, I really tried to buy on aliexpress
<hramrach> but htey did not acept my card
<hramrach> so terminated my account, meh
<JohnDoe_71Rus> hramrach: i bough this
ykchavan has quit [Quit: Leaving]
<oliv3r> tzafrir: tom cubie who does the shop is also the creator fo the cubieboard; personally i'd try getting a olimexino-a20-micro from europe here; big boon, it's OSH
<JohnDoe_71Rus> sorry. bad translate
<JohnDoe_71Rus> I bought *
<tzafrir> oliv3r, yes, those were exactly the things I thought
<arokux> OSH?
<oliv3r> Open Source Hardware
<tzafrir> I gather that the olimexino-a20 does not come with a usb debug cable, right?
<oliv3r> tzafrir: they offer it thoguh
<arokux> oliv3r, olimexino has open PCB layout?!
<oliv3r> arokux: open everyuthing yes
<oliv3r> arokux: check their github under the section 'hardware'
<oliv3r> board files, schematic files, even pdf's of those files are there
<arokux> oliv3r, hm.. why everybody rushes to buy cubies then??? (including me)
<oliv3r> cubie was earlier
<oliv3r> and cubietch is former AW employes?
_enrico_ has joined #linux-sunxi
<arokux> hm.. but hacking is more difficult if pcb layout is unknown.. (I was told this here once)
<arokux> so for us, olimex hardware is better?
<oliv3r> actually, es
<oliv3r> yes
<oliv3r> and has more pins broken out too :)
<oliv3r> double SD cards
<oliv3r> it's a really nice board
<arokux> oliv3r, but no wlan adapter :(
<arokux> so cannot be used as wifi-router
<hramrach> they have wifi option
<JohnDoe_71Rus> cubieboard does not to
<hramrach> plus you can aff as many USB WiFi adapters as you like to anything with USB
<arokux> cubietruck will have
jemk has joined #linux-sunxi
<hramrach> the fact that we still don't have schematic of cubieboard is lame
<hramrach> tom released some preliminary version of the schematic
<hramrach> but the updates that were made during hw debugging are not in that schematic
<arokux> I want some sunxi board that would beat price of RPi...
geecko_ has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<rm> arokux, MK802 is $35 now
<rm> this is much cheaper than the RPi
<rm> $35 <- with free shipping
<oliv3r> from DX?
<arokux> rm, I'd like something hackable..
<arokux> UART connector, pcb layout etc.
<rm> oliv3r, Aliexpress
<arokux> without soldering
<oliv3r> ok cool; i have a DX account, so should check what's avialable there
<rm> the main caveat is to not end up getting an A10S
<oliv3r> i still wnat an M5; but i may wait for whatever comes with A40 when it comes :)
<rm> but even then afaik it can be made to work...
<oliv3r> since a20 may be a little short to do XBMC really well
shineworld has quit [Remote host closed the connection]
<oliv3r> we have a10s support atm i belive
<arokux> guys, why not create our own board, with open everything?
<oliv3r> arokux: you mean, like the olimexino?
<oliv3r> because we don't have to; a) it's way to expensive, b) it exists
<mnemoc> arokux: olinuxino boards are 100% OSHW
<mnemoc> they have a13, a10s, a10 and a20 models
<arokux> but expensive
<mnemoc> that's life
<mnemoc> doing oshw isn't cheap
<arokux> well, as cheap as to do closed source
<mnemoc> but you can take their designs and make your own, and then go to a children-driven factory in somewhere to get your board build cheaper
<arokux> mnemoc, you think children can assemble pcb boards? :)
<arokux> and: why it wasn't done already?
<oliv3r> so if you do it yourself, using their designs probably, since we can't do ddr3 layout ourselves, how much do you think that will cost?
<oliv3r> I sincierly doubt you can make it cheaper
<arokux> yes, one board won't be cheaper
<mnemoc> arokux: olimex launched A20 based SOMs you can hook to a cheap custom 2-layer i/o board
<mnemoc> 35E
<oliv3r> arokux: and olimex is doing the 'many board'
<mnemoc> 80 for the full kit
<arokux> mnemoc, EUR 35 / 1000 pcs
<oliv3r> flash-less a20 boards are 45 Euro?
<oliv3r> how is that expensive
<oliv3r> with flash is 55Euro
<oliv3r> (from the top of my head)
<oliv3r> it's not even that much more then a pi, and you get so much more
<mnemoc> even oliv3r who doesn't get enough money to eat daily find it cheap
<arokux> oliv3r, no, flash-less is 55 EUR, with flash 65 EUR
<oliv3r> close enough
<arokux> my cheap definition is 30-40 EUR :)
<eagles0513875> hi guys
<rm> and likely 20 EUR shipping
<arokux> Shipping - 14.70 EUR
deasy has joined #linux-sunxi
<arokux> oh wait! it was without a tax!
<arokux> VAT 13.92 EUR
<arokux> total: 83.52 EUR - for a flash-less A20
<arokux> oliv3r, ^
<oliv3r> it's without tax on their site?
<oliv3r> that's unlikly
<arokux> well, I've just checked
<oliv3r> crazy :)
<oliv3r> find a local reseller :p
<oliv3r> they do have resellers :)
<arokux> I've got my Mele ages ago for 100 EUR with case, usb hub, wlan...
<oliv3r> some guy actually got it in britain for cheaper
wens has joined #linux-sunxi
rellla has joined #linux-sunxi
<arokux> oliv3r, already checking
<Turl> oliv3r: A10S ain't that bad :) my OlinuXino is awesome
<rm> distributors typically jack up the price by 1.5 to 2x
<rm> ordering directly is better
<arokux> with VAT = 65 EUR, on olimex website: 72.17 EUR
<arokux> and shipping is 3 EUR here, on olimex - 15 EUR
<arokux> a lot of hidden costs...
wingrime has quit [Ping timeout: 276 seconds]
<rm> meh
<rm> I will just keep buying from Aliexpress
<rm> free shipping, and I pay no VAT
<Turl> rm: are you still using that mk802 router?
<oliv3r> Turl: why would a10s be bad?
<arokux> rm, in that case customs should be paid ;)
<rm> I am not paying customs
<arokux> rm, where are you from?
<oliv3r> rm: if you don't care about OSH; then sure
<rm> in Russia you can receive up to 1000 EUR/month of goods tax-free
<oliv3r> otherwise, support olimex and buy from them(or a distributer)
<rm> Turl, I migrated that into a VM on my large server
<rm> and it was not MK802 but a CB later on
<mnemoc> in .de it's merly 40E :(
<mnemoc> merely*
<Turl> 07:05 rm> [06:40:27] the main caveat is to not end up getting an A10S
<Turl> oliv3r: ah, it was rm, not you
<arokux> mnemoc, what is 40?
<rm> well
<arokux> in de you need to pay VAT if price is between 22 and 150 EUR
<Turl> mnemoc: don't complain ;) 25USD/y here :p
<rm> I bet majority of those won't work on an A10S out of the box
<Turl> rm: fedora does
arete74_ has quit [Read error: Connection reset by peer]
arete74 has joined #linux-sunxi
<oliv3r> arokux: 2 x 20
<oliv3r> arokux: 2 short of the answer to all questions, 4 x 10, 41 -1; what is the question?
<mnemoc> Turl: :D
<arokux> oliv3r, what? :)
<arokux> oliv3r, I'd love to support olimex, but it is too expensive to do it. except of this I'm already supporting them by hacking and answering questions here sometimes...
<oliv3r> arokux: i think mnemoc ment 40E free import a month tax-free
<oliv3r> arokux: haha, well then maybe you'll get a next version as a developer from them ;)
<arokux> I think mnemoc isn't right..
<oliv3r> i think it's 44 E for NL
<oliv3r> so i think he's very muhc rgiht
<oliv3r> holy shit, on the dutch distributer site, the olimex is expenis,ve but even a rpi-B goes for 46E
<Turl> oliv3r: 90USD for a rpi-B on a local distributor here :(
<jemk> hramrach: do you still have problems with vdpau crashing?
<arokux> has it for 40 EUR with shipping
<Turl> oliv3r: and they carry cubieboards :P for 150US$
<arokux> < 22 EUR tax free
<oliv3r> yeah it's even worse
<oliv3r> i think it was 24E here actually
<oliv3r> BUT! within the EU there's no limit
<mnemoc> sure, you just pay ~20% of VAT for everything :p
<arokux> yes :( and shipping...
<oliv3r> i order tons of ebay/dx stuff from china too, so i'm no saint, but technically it's fraud
<oliv3r> you are supposed to pay taxes on those products :)
<arokux> oliv3r, well, everything goes through customs office where I live, so you cannot avoid paying VAT ;p
<Turl> oliv3r: IRS is gonna go after you now that you said that :p
<oliv3r> arokux: oh i verry often dont have to
<oliv3r> heh :0
oliv3r has quit [Quit: Changing server]
oliv3r has joined #linux-sunxi
massi has quit [Read error: Connection reset by peer]
_enrico_ has quit [Read error: Connection reset by peer]
massi has joined #linux-sunxi
jemk has quit [Ping timeout: 276 seconds]
<arokux> the problem with the board is that they are not mainstream products
<arokux> a bunch of hackers buy them
<mnemoc> that's the point of SoMs
<mnemoc> you are supposed to build custom products with them, cheapily
<arokux> Mele M5 with casing, remote and wlan is 65 EUR...
<arokux> and 8Gb NAND!
<oliv3r> cheap protoype boards or cheap boards to put in homedesigned motherboards
<oliv3r> arokux: without taxes :p
<mnemoc> CE vs. devboards
<arokux> CE?
<oliv3r> mele m5 only 65 now? i saw it for 88 at DX
<oliv3r> Consumer Electronics
<oliv3r> ah 88 USD vs 65 Euro's
<arokux> oliv3r, 65 EUR = 88 USD
<arokux> yes
<oliv3r> yeah that's the one i was looking at
<oliv3r> i want to see the inside
<oliv3r> if VGA port can be connected easily to the PCB
<mnemoc> arokux: here in general USD is the default currency, unless the E is appended
<mnemoc> if you tell 65 people will assume dollars
<arokux> ok, thanks
<oliv3r> which is funny; since most of us are non-US :)
<mnemoc> true :D
<arokux> we should ask smb with mele m5 to take pictures of the board
<mnemoc> and wiki it
<arokux> yes, actually it is strange it's not there yet
hipboi_ has joined #linux-sunxi
<oliv3r> who has mele m5?
massi has quit [Ping timeout: 264 seconds]
<arokux> and now tell me the board that will beat mele m5 price even ignoring casing and remote
hipboi has quit [Ping timeout: 256 seconds]
rz2k has quit [Read error: Connection reset by peer]
<arokux> jelly-home, and?
<oliv3r> oh wow that's old stuffs :)
<oliv3r> but yeah; now we need the same for mele m5
<arokux> but without SATA
<arokux> I'm wondering why everybody needs sata if there is usb?
hipboi_ has quit [Remote host closed the connection]
<oliv3r> usb is hella slow and buggy as hell
<oliv3r> :p
<oliv3r> and sata is 'nice to have'
<rm> SATA is 3-4x faster than USB
<rm> also the latency is perhaps an order of magnitude less
<rm> and it should consume less CPU
<arokux> are there external hard drives with SATA?
<rm> for extrenal drives it's called eSATA
<rm> generally the same thing, a bit different plug
<JohnDoe_71Rus> but 1 sata not enouth to nas
geecko__ has joined #linux-sunxi
<leviathanch_> JohnDoe_71Rus: unfortunately the A20 has only one SATA controller
<leviathanch_> if you need more you have to add a USB-SATA-Bridge
<leviathanch_> (or multiple USB-SATA bridges)
<rm> JohnDoe_71Rus, you can get that Mele M5
<rm> plug in a SATA SSD on top
<rm> and that's a pretty nice ARM desktop machine
<JohnDoe_71Rus> and limited bus usb
geecko__ has quit [Client Quit]
geecko_ has quit [Quit: Quitte]
<geecko> what the heck, i had three instances of HexChat launched \o/
<oliv3r> exciting times!
<oliv3r> what's more intersting, is if the a10/a20 supports port multiplier :)
<rm> * geecko__ has quit (Client Quit)
<rm> * geecko_ has quit (Quit: Quitte)
<rm> <
<rm> some in french, too
<oliv3r> rm: eSata also does things with the signal, makes it more robust and a bit more power or something
<rm> noooo it dooooesn't~
<rm> given there are passive convertors SATA<>eSATA
<oliv3r> the signaling is compatible
<oliv3r> yes
<oliv3r> but there's also bios options to make a sata port behave as an eSATA port :)
<oliv3r> for good reason
<rm> doesn't that just set a "detachable" flag somewhere
Black_Horseman has quit [Remote host closed the connection]
ganbold_ has joined #linux-sunxi
<oliv3r> i think it does a little more, but in any case, its fully compatible and interchangeable
hansg has joined #linux-sunxi
<arokux> leviathanch_, please remember me what is your board :)
<leviathanch_> arokux: cubieboard2
<leviathanch_> A20
<arokux> ok
jemk has joined #linux-sunxi
<leviathanch_> Turl: clk_ignore_unused worked
_enrico_ has joined #linux-sunxi
<arokux> leviathanch_, what does this option do?
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier.]
\\Mr_C\\ has quit []
<oliv3r> Turl: what is the best string to grep for to see how the emac_phy gets handled in code?
<leviathanch_> arokux: it prevents unused clocks from being reinitialized
<leviathanch_> arokux: which is a neckbreaker for a system
<leviathanch_> in case not all RAM clocks and cpu clocks are getting claimed in the right order by the appropriate device
<arokux> hm.. maybe I should try it too, I have a problem with my usb host for mainline, do not know what to try next..
_enrico_ has quit [Ping timeout: 246 seconds]
_enrico_ has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
boycottg00gle has joined #linux-sunxi
<arokux> leviathanch_, where are you from in Switzerland?
<leviathanch_> Zürich
<leviathanch_> arokux: Zürich
<leviathanch_> arokux: ok
<leviathanch_> arokux: you are from germany, right?
<leviathanch_> the good thing about mmc is
<leviathanch_> you can debug it with an oscilloscope partially
<leviathanch_> :-)
<leviathanch_> by seeing if the data and clk pins are functioning properly
<arokux> leviathanch_, why have you picked mmc, btw?
<leviathanch_> since it's one of the firts things you'd expect a tablet to do
<leviathanch_> reading a system from mmc
<arokux> leviathanch_, and wireless?
<mnemoc> Zürich... the city of the god called niklaus wirth
* mnemoc envies leviathanch_
VicenteH has quit [Ping timeout: 240 seconds]
ZetaNeta has quit [Ping timeout: 264 seconds]
<mnemoc> algorithms + data structures = programs was one of my first books
rellla has quit [Quit: Nettalk6 -]
hipboi has joined #linux-sunxi
<oliv3r> Turl: irq ping
hansg has quit [Quit: Leaving]
boycottg00gle has quit [Remote host closed the connection]
<arokux> any idea why sun4i_defconfig includes "Support for large (2TB+) block devices and files"
* arokux wonders if it should be disabled to speed up compilation
akaizen has quit [Read error: Connection reset by peer]
<atsampson> arokux: why wouldn't you want that enabled? 2TB+ hard disks are common now, and plenty of sun4i boards have SATA...
akaizen has joined #linux-sunxi
oliv3r has quit [Ping timeout: 276 seconds]
oliv3r has joined #linux-sunxi
Soru has quit [Ping timeout: 248 seconds]
<mnemoc> defconfigs are only a reference for testing code changes and as starting point for your custom thing
<mnemoc> not "the ultimate .config"
<mnemoc> but patches are welcomed
<mnemoc> specially to make them consistent
<Turl> oliv3r sup
<Turl> on mobile atm
<arokux> mnemoc, so you are for 2TB+ or against them? :)
<mnemoc> i don't vote
<mnemoc> it's community choice
<mnemoc> more complete defconfigs let me test more
<mnemoc> (build time for each defconfig)
<mnemoc> but i don't vote
<arokux> ok :)
<Turl> i dont think you would gain much disabling that
<Turl> oliv3r, look for sun4i-mdio file
<arokux> Turl, isn't it same for any option? but if 10 options are found this is already speed up :)
<oliv3r> Turl: do we have any drivers in ML now that use IRQ's on sunxi?
<oliv3r> Turl: i only found emac so far
<oliv3r> which looked a little messy :p
<Turl> oliv3r, i2c? dunno
<oliv3r> oh sun4i-mdio is different from sun4i-emac :)
<Turl> look for interrupts on dt
<oliv3r> Turl: yeah but i2c never got merged; we merged sunxi ontop of the marvell one
massi has joined #linux-sunxi
<oliv3r> Turl: well the real thing is
<oliv3r> why not use: platform_get_resource(pdev, IORESOURCE_IRQ, 0);
<oliv3r> but instead i see a lot platform_get_irq(pdev, 0);
<oliv3r> but there's also irq_of_parse_and_map(&pdev->dev.of_node, 0);
<Turl> shorthand maybe?
<oliv3r> and of course always devm_request_irq()
<oliv3r> so ther's so many differnt ways of doing it, i can't clearly see what the recommende dway with DT is
<arokux> I've just found drivers/misc/sunxi-dbgreg.c maybe it could be useful for you guys
<arokux> leviathanch_, ^
<Turl> remember that of fujctions usually rescan dt
<mripard> see what it's doing ? :)
<Turl> vs prescanned plat devices
<oliv3r> mripard: ah hah!
<oliv3r> so platform_get_irq it is, so why use plaform_get_irq, and not platform_get_mem (or whatever its equivilent is)
<Turl> of_iomap?
<Turl> :p
<oliv3r> so there is no platform_get_mem :p
<mripard> oliv3r: because there's no such thing as platform_get_mem :)
<oliv3r> hehehe ok
<arokux> is there a way to build everything into an image, not modules without going through all the options and changing M to Y?
<Turl> mripard did you see my msg about sid
<Turl> ?
<mripard> Turl: 3.12 vs 3.13 ?
<Turl> yes
<mripard> yeah, the point is, the merging will be a pita.
<mripard> since oliv3r patches file that are not in greg's tree
<Turl> ah
<mripard> that would require greg to put is branch on top of mine, etc.
<Turl> maybe ojn can take it then
<mripard> the ARM merge window is pretty much closed
<mripard> he already postponed "real pull requests to 3.13
<Turl> for rc1?
<Turl> hm
<oliv3r> so with greg pushing it in 3.12-rc1; what will that mean, only the driver, but not the dt will land?
<Turl> yes
<Turl> so no working sid ootb til .13
<oliv3r> awaw
<oliv3r> well atleast the driver will be there, so that's good
<Turl> yeah I guess
<mripard> and that gkh has priviledges that we don't have :)
<Turl> gkh can ship all the things :)
<oliv3r> yeah go gkh go!
<Turl> unreliable wifi is unreliable
<Turl> totp doesnt make it fun to reattach to irssi
<oliv3r> hometime
<arokux> does sunxi need it?
stulluk has joined #linux-sunxi
<stulluk> Is there a way to enable
atiti has joined #linux-sunxi
<rah> can you do fastboot with allwinner devices?
ykchavan has joined #linux-sunxi
ykchavan has quit [Client Quit]
atiti has quit [Ping timeout: 240 seconds]
stulluk has quit [Ping timeout: 276 seconds]
panda84kde has quit [Quit: Konversation terminated!]
ykchavan has joined #linux-sunxi
ykchavan has quit [Ping timeout: 248 seconds]
stulluk has joined #linux-sunxi
naobsd has joined #linux-sunxi
wingrime has joined #linux-sunxi
VicenteH has joined #linux-sunxi
hipboi has quit [Remote host closed the connection]
wingrime has quit [Ping timeout: 245 seconds]
ZetaNeta has joined #linux-sunxi
<rah> how can I extract boot0 and boot1 binary images from a image intended for flashing by LiveSuit?
wingrime has joined #linux-sunxi
ykchavan has joined #linux-sunxi
massi has quit [Remote host closed the connection]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
n01 has quit [Remote host closed the connection]
_enrico_ has quit [Quit: Bye]
dwilkins has joined #linux-sunxi
cyp has quit [Read error: Connection reset by peer]
CoSM0 has quit []
\\Mr_C\\ has quit []
stulluk has quit [Ping timeout: 245 seconds]
ykchavan has quit [Read error: Connection reset by peer]
\\Mr_C\\ has joined #linux-sunxi
wingrime has quit [Ping timeout: 246 seconds]
wingrime has joined #linux-sunxi
stulluk has joined #linux-sunxi
<ojn> mripard: Turl: Hard to track what you're talking about in the backlog (and i'm lazy :). What's up?
BJfreeman has quit [Quit: had a good time]
<ojn> mripard: ah. It's not really about privileges as much as the mood and availability of the maintainer at any given time, to be honest; adding new drivers _is_ generally ok later in the release cycle, but if it's usually done on best-effort basis.
<ojn> luckily, the dts and binding change can go in completely independent of the driver. it could even have been merged before the driver (if the binding is agreed upon and reviewed and intended to keep stable)
<leviathanch_> Turl: where can I find your patch for automagic reparenting?
<wingrime> ssvb: ping
<ssvb> wingrime: pong
<wingrime> ssvb: I find some workaround for disp shake
<wingrime> ssvb: try enable one scaler
<wingrime> ssvb: I think prbolem in syncronization between layers
<wingrime> ssvb: if some have no scaler , they quick
<wingrime> ssvb: if layer have scaler it little slower
<wingrime> ssvb: so I think it sync wait make it shake
<wingrime> ssvb: i enabled scaler in fex, it fixed issue
<wingrime> only one scaler
<Turl> ojn: I guess you read the backlog by now :)
<leviathanch_> Turl: have you pushed it already?
<leviathanch_> since I really need to cherrypick it
<Turl> ojn: the binding is pretty simple, just a compatible and reg
<Turl> leviathanch_: not yet, but I can push my draft copy on a tag if you really need it
<leviathanch_> Turl: yes
<leviathanch_> I need it
<leviathanch_> as far as it is already doing something usable
<leviathanch_> see it positive: it will then be tested on A20 as well ;-)
<Turl> leviathanch_: ok, I'll push it, give me 10m
<leviathanch_> Turl: ok
wingrime has quit [Read error: Operation timed out]
stulluk has quit [Ping timeout: 248 seconds]
dwilkins has quit [Ping timeout: 248 seconds]
dwilkins has joined #linux-sunxi
jemk has quit [Ping timeout: 246 seconds]
wingrime has joined #linux-sunxi
<Turl> leviathanch_: usual repo, tag `clk-sunxi-draft-for-leviathanch`
<Turl> btw, as it worked with the ignore unused clocks boot flag, would you mind pasting clk/clk_summary from debugfs?
fredy has quit [Excess Flood]
<mnemoc> Turl: hi, you told me there were two patches in stage to squash before merging to non-stage.... do you remmeber which were day?
fredy has joined #linux-sunxi
<mnemoc> Turl: I would like to finally merge this thing
<Turl> mnemoc: sure it was me? I don't recall doing that
<Turl> mnemoc: might have been arokux
<ssvb> mnemoc: maybe the mali regression?
<mnemoc> it was before merging the first a20 bundle from hansg
<Turl> ssvb: I thought hans picked that?
<Turl> ssvb: he picked it up as is though, #error et al :p
<leviathanch_> Turl: I'm going to boot it
<mnemoc> are there critical issues open in the current stage-3.4? or can I just merge it?
<Turl> mnemoc: arokux was complaining that usb was broken
<Turl> due to one of his own patches
<Turl> was asking for a revert, check the ML
<mnemoc> reverted one he asked here on IRC
<mnemoc> last night
<Turl> ok then
<mnemoc> i suppose it's the same
<ssvb> mnemoc: mali is broken on sun4i with stage/sunxi-3.4
* mnemoc breaths and take the courage to download the >3k pending mails over 3G
<mnemoc> ssvb: :(
<mnemoc> ssvb: known fix?
<mnemoc> or known breaker?
<mnemoc> to leave out of the merge
<Turl> leviathanch_: that's with or without the ignore?
<leviathanch_> with the ignore
<leviathanch_> and yes
<Turl> leviathanch_: old or new code?
<leviathanch_> ?
<leviathanch_> you mean with your patch?
<leviathanch_> not ye
<leviathanch_> *yet
<Turl> patches*
<Turl> you need to pick up a bunch of them :)
<leviathanch_> just doing a fetch
<leviathanch_> :-)
wingrime has quit [Ping timeout: 245 seconds]
wingrime has joined #linux-sunxi
wingrime_ has joined #linux-sunxi
<ssvb> mnemoc: I guess hansg mentioned this pull request -
<mnemoc> ssvb: thanks
techn_ has joined #linux-sunxi
<mnemoc> ssvb: fetching
techn__ has quit [Ping timeout: 260 seconds]
iamfrankenstein has quit [Ping timeout: 256 seconds]
jemk has joined #linux-sunxi
jemk_ has joined #linux-sunxi
jemk has quit [Disconnected by services]
jemk_ is now known as jemk
Olaff has joined #linux-sunxi
robb83 has joined #linux-sunxi
<mnemoc> building
<oliv3r> jobs done?
<wingrime_> mnemoc: are you moving stage ?
<oliv3r> hmm, i have 140 'kworker/u12' spawned
<mnemoc> imerged hansg's pull request
<mnemoc> i just merged hansg's pull request
<mnemoc> but test building it now
<mnemoc> would like to only add critical patches to stage before the next merge to non-stage
<wingrime_> mnemoc: than add my fix for usb and for scaler
<wingrime_> mnemoc: otg realy regression
<wingrime_> ssvb: please make layer request only on demand
<wingrime_> ssvb: I can't test libvpdau with one scalar
<mnemoc> wingrime_: paste me the subjects please
<mnemoc> i'll add them now
<geecko> toolchain 4.7 -> "android-toolchain-eabi-4.8-2013.07-x86.tar.bz2"
<geecko> still not fixed after a week
<mnemoc> we are not linaro
<wingrime_> [PATCH] sunxi:disp: Fix problems with negative overlay position
<wingrime_> [PATCH] sunxi:usb: Fix otg usb mode selection
<mnemoc> wingrime_: ssvb seems to have nacked it....
<wingrime_> mnemoc: ssvb sayed that we can remove "x" coordinate fix
<wingrime_> mnemoc: bug actualy only with "y" negative coordinate
<wingrime_> mnemoc: you can remove "x" coordinate workaroud form patch when merge
<mnemoc> and the other nacked by techn :(
<mnemoc> please pull an ack from them if they now agree
<wingrime_> ssvb: ^^^
<ssvb> wingrime: even with only "y" coordinate, it is still not entirely correct
<wingrime_> mnemoc: don't remeber anything about techn
<wingrime_> ssvb: I have no other idea now
<ssvb> wingrime: you are forcefully changing the overlay position, and this is not what the userspace caller expects
<wingrime_> ssvb: there is chance that it hw bug
eebrah has quit [Quit: .]
<ssvb> sure
<wingrime_> ssvb: yes, but I can't do anything else
<ssvb> but a nicer workaround can be probably found instead of just chopping it off
<ssvb> I can workaround the problem in the userspace for XV
<geecko> mnemoc, did i said so? :)
<ssvb> wingrime: just give me a day or so
<wingrime_> ssvb: also do you tryed enable one scaler in fex and check for screen shake in hi resolution?
<wingrime_> mnemoc: it looks like hw bug
<wingrime_> mnemoc: also I not remeber anyting wrong for second patch...
<mnemoc> see the thread
<mnemoc> one response from hansg and two from techn_
<wingrime_> mnemoc: not techn
<wingrime_> mnemoc: also
<wingrime_> mnemoc: Hmm.. sorry. This diff fixes host mode only. OTG seems to be having other problems.
<mnemoc> jari is techn_
<wingrime_> mnemoc: he writed , that his claim was wrong
<wingrime_> mnemoc: so I not see any problems for this patch
<techn_> wingrime_: Yes. Claim was wrong. Host mode is working ok.. and according hansg OTG mode too
<techn_> .. without your and my CONFIG_USB_PORT_POWER_MANAGEMENT patches
<wingrime_> techn_: It defentely fixed otg for me
<mnemoc> still sounds like a request for a v2 in both cases
<wingrime_> techn_: where hasng sayed that it works for him?
<wingrime_> techn_: he only writed >>> I believe this patch is against 3.0, correct ?
<techn_> ..
<techn_> "Besides the above patch being wrong, AFAIK otg mode works fine, I
<techn_> always build the Fedora kernels with OTG mode at the Kconfig level,
<techn_> and most fex files also use:"
<wingrime_> techn_: ...
<wingrime_> techn_: thats impossible, UDC0 not even builds in kernel
<wingrime_> techn_: impossible even select it
<techn_> It selects here
<techn_> Might be that it wont select if it is build as module
eebrah has joined #linux-sunxi
<wingrime_> techn_: do
eebrah is now known as Guest96066
<wingrime_> techn_: ,config | grep USB_SW_SUNXI_UDC0
<wingrime_> .config | grep USB_SW_SUNXI_UDC0
<wingrime_> cat .config | grep USB_SW_SUNXI_UDC0
<mnemoc> cat | grep looks kind of funny
<techn_> mnemoc: I usually do that :)
<techn_> but anyway.. it is selected on me CONFIG_USB_SW_SUNXI_UDC0=y
<mnemoc> uh
<techn_> easy to cat or grep according situation :p
<wingrime_> techn_: ok, try USB_SW_SUNXI_USB0_OTG
<leviathanch_> Turl: if think there is something broken with the branch you have forked from
<leviathanch_> it won't load initrd from u-boot
<wingrime_> techn_: ?
<techn_> wingrime_:
<techn_> wingrime_: I'll paste new
<wingrime_> techn_: can find USB_SW_SUNXI_USB0_OTG
<techn_> wingrime_:
<wingrime_> techn_: this is 3.4 stage?
<techn_> there is full .config
<techn_> wingrime yes
<wingrime_> techn_: ok, try cp sun5is
<wingrime_> sun5i def config
<techn_> ?
<wingrime_> than run make menuconfig, and exit with save
<techn_> done that
<wingrime_> can paste me config ?
<techn_> it's same
<wingrime_> not sure
Soru has joined #linux-sunxi
<wingrime_> techn_: I try again myself
<wingrime_> techn_: you config can be old enought to not have this issue
<wingrime_> techn_: but I remeber that I have problem
<techn_> default config might need some update
<wingrime_> techn_: bug can't hide by himeslf
<libv> wingrime_: about using the mali for everything...
<libv> wingrime_: sure, the mali will make any calculations you can dream of.
<libv> wingrime_: but that doesn't mean that it is the most time or energy efficient way of doign things
<libv> colourspace conversion is best left to the display engine
<libv> idct decoding is best left to the media decoder
<oliv3r> and scaling?
<libv> also display
<libv> that's hardware that's designed for this specific purpose
<libv> sure, someone needs to go write the code, but it pays off big time
<libv> usually, if the hardware is known good, bringing up an overlay which does colour space conversion for you, is pretty trivial if you already have everything else wired up correctly
<leviathanch_> Turl: linux-next boots :->
<leviathanch_> from the official repo
<wingrime_> techn_: bug realy exsist
<wingrime_> techn_: try go to menuconfig
<techn_> wingrime_: yes?
<wingrime_> techn_: to [*] SUNXI USB2.0 Dual Role Controller support --->
Soru has quit [Read error: No route to host]
<wingrime_> techn_: and try find otg mode
<techn_> wingrime_: It's selected
<wingrime_> techn_: (X) host only support
<wingrime_> techn_: see?
<techn_> wingrime_:
<wingrime_> techn_: how?
<techn_> Depends on: USB_SUPPORT [=y] && USB_SW_SUNXI_USB [=y] && USB_SW_SUNXI_USB_MANAGER [=y]
<techn_> make sure that your dependencies are met
dwilkins has quit [Ping timeout: 256 seconds]
hramrach has quit [Ping timeout: 240 seconds]
<wingrime_> techn_: please , git log
<techn_> Depends on: <choice> && USB_SW_SUNXI_UDC0 [=y]
Soru_ has joined #linux-sunxi
<wingrime_> techn_: I just showed default config
<wingrime_> alex@laptop-lenovo:/media/archive1/sunxi/linux-sunxi$ cp arch/arm/configs/sun5i_defconfig .config
<wingrime_> alex@laptop-lenovo:/media/archive1/sunxi/linux-sunxi$ make menuconfig ARCH=arm
<wingrime_> techn_: wtf
<wingrime_> techn_: so, or you have some old config , or you use some other branch
<wingrime_> techn_: or I don't know what a hell is it
<wingrime_> mnemoc: can you try this ?
iamfrankenstein has joined #linux-sunxi
<wingrime_> techn_: or you use old branch
hramrach has joined #linux-sunxi
dwilkins has joined #linux-sunxi
<wingrime_> techn_: so?
mirc_243421 has joined #linux-sunxi
mirc_243421 is now known as Legitsu
<hno> wingrime, what are you trying to do?
<wingrime_> hno: prove that my patch realy needed
<hno> wingrime, and your patch does?
<wingrime_> hno: fix otg-only building and fix kconfig to make it workable
<wingrime_> hno: I can't get working otg cable with my tablet after patch
<wingrime_> *after usb merge
<wingrime_> techn_: ?
<techn_> wingrime_: select USB_SW_SUNXI_UDC0 , then you'll see OTG mode avalable
<wingrime_> techn_: my patch change "depend on" to "selects" also it fixes build for otg mode only
<wingrime_> techn_: try build otg only
<wingrime_> techn_: it will defently fall
<techn_> otg is both UDC and HDC
<leviathanch_> Turl: clk_set_rate leads to a crash in the background
<leviathanch_> Turl: clk_round_rate already sets the clock!
<leviathanch_> something is a bit wrong here
iamfrankenstein has quit [Ping timeout: 240 seconds]
paulk-aldrin has joined #linux-sunxi
<techn_> UDC is device mode, HDC host, OTG==UDC+HCD
<wingrime_> techn_: try build "device only"
<wingrime_> mnemoc: ping
uwe_ has quit [Remote host closed the connection]
<wingrime_> techn_: I still think that my patch needed for make it more easy to select all
<wingrime_> mnemoc: ^
<wingrime_> mnemoc: yeax it possible make otg without patch but it non trivial
<Legitsu> might I ask whats wrong with simply enabling it in a defconfig
<Legitsu> =/
<oliv3r> do we have the 3.4 lycheekernel source anywhere up?
<oliv3r> (on a git/lxr so i can browse it)
<Turl> leviathanch_: the base branch is clk-next+arm dt pull, if you're using A20 it's likely it won't boot or it'll stall, needs more pinctrl and maybe gic patches
<Turl> leviathanch_: clk_round_rate should just compute a suitable rate
<Turl> leviathanch_: setting which clock crashes?
<leviathanch_> hard to say
<leviathanch_> since it just stops
<leviathanch_> but it crashes when I access the mod0 mmc0 clock
fredy has quit [Excess Flood]
<Turl> leviathanch_: it's probably reclocking pll5
<Turl> reclocking ram just because is bad for your board :)
<leviathanch_> I know ;-)
<leviathanch_> that's why I had masked pll5 in clock selection
<leviathanch_> when I did it myself
fredy has joined #linux-sunxi
<leviathanch_> couldn't you skip the set rate function if the name is or contains pll5?
<leviathanch_> or just don't use it as a parent?
<Turl> 1) is an ugly hack, and there's legit reasons for changing its rate (although probably without the framework)
<leviathanch_> ok
<Turl> 2) means having a fake hw description
<leviathanch_> ok
<leviathanch_> maybe I should just throw pll5 out of the list in the dt?
<Turl> that's 2)
<Turl> we could improve the reparenting code to not consider pll5 maybe
<leviathanch_> Turl: yes
<Turl> what's wrong with the uboot mmc clock btw?
<leviathanch_> ?
<Turl> ie why do you need a diff rate?
<leviathanch_> osc24M is slow
<Turl> if you're setting the same rate the code should be a noop
<Turl> ah, uboot uses it on osc24?
<leviathanch_> there are _many_ different rates
<leviathanch_> uhm
<leviathanch_> moment pls
<Turl> my board was on PLL5 @ 100Mhz I think
<leviathanch_> uhm
<leviathanch_> yes
<leviathanch_> sorry
<leviathanch_> that's correct
<leviathanch_> but mmc_host starts down at 400kHz
<leviathanch_> for some strange reason
<leviathanch_> ah
<leviathanch_> the reason is there as a table-comment
jemk has quit [Ping timeout: 245 seconds]
<leviathanch_> 400kHz is the slowest speed an SD3.0 card may support
<Turl> 100MHz > 400kHz
<leviathanch_> yes
<leviathanch_> _but_ the host driver of Linux wants to start at the deepest speed rate
<leviathanch_> and then ask how fast the card can become
<leviathanch_> because a 400kHz card can't say: "It's too fast" if it's asked with 100MHz
<leviathanch_> but a 100MHz card can say: It's too slow. Boring!
<leviathanch_> ;-)
akaizen has quit [Remote host closed the connection]
<wingrime_> mnemoc: ping
<leviathanch_> Turl: I could also use __ functions and select pll6 in case the frequency needs to be changed
<leviathanch_> Turl: given you have finally implemented reparenting
<leviathanch_> which shouldn't be hard
<leviathanch_> you have a limited number of possible clocks
<leviathanch_> and inputs for them
<oliv3r> Turl: i can't figure out how you map a regulator to a driver :S but the name suggests it has no reference to emac_3v3 from the DT, so how do they get together?
<leviathanch_> 00 OSC, 01, 10, 11
<leviathanch_> Turl: 00 OSC, 01, 10, 11
<leviathanch_> so checking pclk->name
<leviathanch_> and looking the mode up in a table
<leviathanch_> should do
<Turl> leviathanch_: can you try this and see if it stops crashing?
<oliv3r> yeah that bit i'm talking about emac_3v3 is the name of the regulator
<Turl> phy-supply = <&regulator>
<Turl> devm_regulator_get(&pdev->dev, "phy");
<oliv3r> so it's "name"-supply?
<oliv3r> strange construct, but I guess that'll have to do :)
<Turl> that's my guess
tm512 has joined #linux-sunxi
wingrime has quit [Ping timeout: 248 seconds]
arokux2 has joined #linux-sunxi
<arokux2> mripard, around?
<arokux2> or maybe smb else, here is the problem:
<Turl> oliv3r: I guess dt code just matches them with the right pdev, and uses s/-supply/ on the key to set the name :p
pirea has joined #linux-sunxi
<leviathanch_> Turl: the patch leads to an even bigger issue
<arokux2> I'm trying to pull in allwinner code without dt now. just set everything manually in registers.
<leviathanch_> obviously the system trys to set an initial clock frequency on pll5
<leviathanch_> by preventing the clock from being changed
<leviathanch_> the system won't even start booting
<leviathanch_> at least it now hangs at "Uncompressing Linux... done, booting the kernel."
FunkyPenguin has quit [Read error: Connection reset by peer]
<arokux2> leviathanch_, fdt_high is set?
<Turl> oliv3r: of_get_regulator on drivers/regulator/core.c
<leviathanch_> arokux: it did just boot before I tried the patch of Turl
<oliv3r> Turl: let me read up on that
<Turl> leviathanch_: nope, linux never touches the pll5 code, it's set by uboot and let alone
<Turl> pll5 freq*
<leviathanch_> wtf then?
<oliv3r> Turl: oh i had that section open, well regulatr_get()
<oliv3r> but exactly, i was expecting someth8ing like that somewhere
FunkyPenguin has joined #linux-sunxi
<oliv3r> ah found it, devm_regulator_get(); calls regulator_get() which calls regulator_dev_lookup, which calls, of_get_regulator :)
<oliv3r> snprintf(prop_name, 32, "%s-supply", supply); and there it is
<tm512> ssvb: why wouldn't there be display support in the mainline kernel at all? would it be too hard to bring it over myself?
<oliv3r> it's always %s-supply
<oliv3r> tm512: it's will be a huge undertaking
<leviathanch_> Turl: something within this patch breaks the initial boot process anyway
<leviathanch_> without the patch it loads just fine
<tm512> it isn't a module?
<oliv3r> tm512: it's a huge horrible mess
<oliv3r> tm512: it's a big hack-job
<oliv3r> tm512: in its current state, it will never ever be accepted in ml
<tm512> I mean, taking my own clone of ml, and merging it in, regardless of whether it's a mess
<Turl> leviathanch_: ah, I derped the variable name, replace clk_name with parent on the strcmp
<Turl> leviathanch_: btw, I'd highly recommend you enable early_printk, debug_ll and pass earlyprintk on the cmdline
<Turl> you'd have seen the crap dereference :)
<ssvb> tm512: this is what is supported in the mainline kernel at the moment -
<oliv3r> tm512: ohh, you wanna have messy drivers in a local git repo; i guess that could work for sunxi once we depreciate 3.4
<ssvb> tm512: as you can see, the display driver has not been done yet
<libv> tm512: are you serious?
<ssvb> tm512: if you want a feature complete system, you can use the sunxi-3.4 kernel based on the vendor code drop
<libv> tm512: you are intending to do linux-from-scratch on a weird NIH libc called musl
<libv> and now you also want to port kernel drivers to mainline along the way as well?
<libv> tm512: am i wrong in assuming that you will also be curing cancer in between?
<oliv3r> er what, waht's musl?
<libv> oliv3r: yet another re-invention of libc
<wingrime_> libv: yeax NIH
<leviathanch_> Turl: ok
<leviathanch_> thrown earlyprintk into cmdline
<arokux2> guys want to do some hacking, do not prevent him from it!!!
<arokux2> we need hackers
<leviathanch_> must have gone lost during the attempts to boot with initrd
<pirea> hy all
<pirea> libv you are still working at lima?
<libv> pirea: yes, why?
<pirea> libv because the code was no changed from 3 months ago
<libv> arokux2: the art of being a hacker is setting tough goals and actually working towards them logically and then achieving those goals, which is quite something different from daydreaming up and endless amount pointless goals
<leviathanch_> Turl: when you autoreparent
<leviathanch_> do you enable the new parent then as well?
<libv> pirea: that's because i haven't done much code in a last 3 months
<libv> pirea: you might want to read my latest blog entry though
<Turl> leviathanch_: that happens when you enable the clock
uwe_ has joined #linux-sunxi
<leviathanch_> it hangs
<leviathanch_> doesn't do anything anymore
<Turl> when does it hang?
<leviathanch_> after the clk_set_rate
<leviathanch_> do I need to disable the clock first?
<Legitsu> nice libv
<Turl> leviathanch_: can you comment out the set_rate and paste clk_summary?
<Turl> no
<libv> i just got textures and mipmaps working as well
<pirea> libv is your blog?
<tm512> libv: wouldn't it be somewhat straightforward to bring the display driver source from 3.4 to mainline, or has that changed so much that it would be an impossible task?
<Legitsu> so the gpu accually does something other then look pretty and produce heat now
<Legitsu> nice
<libv> now for elements buffers and then cleanup and another big push
<libv> tm512: why don't you give it a go
<wingrime_> libv: nice, when it will become real driver
<wingrime_> ?
<libv> wingrime_: you do know that i had ioquake3 running at fosdem, right?
pitillo has quit [Ping timeout: 245 seconds]
<Turl> wingrime_: I think he means "ETA for mesa et al?"
<libv> wingrime_: which means that i had solid knowledge on how to do things like texturing and mipmapping
<Turl> err, libv ^
<libv> wingrime_: so why would i be happy and proud 7 months on about having texturing and mipmapping working?
<Turl> leviathanch_: ok, now, what clock are you enabling that makes it explode?
<wingrime_> libv: ?
<libv> wingrime_: alternatively, you could read
pitillo has joined #linux-sunxi
<leviathanch_> Turl: mmc0
<wingrime_> libv: sdl2
<Turl> leviathanch_: but it's already enabled according to what you just pasted?
<wingrime_> Turl: this is a first thing I want
<tm512> libv: ah, fun, looks like I would have to bring the mmc driver over too, or that wouldn't work very well at all
<Turl> tm512: leviathanch_ is working on that
<leviathanch_> yes
<libv> wingrime_: sdl2 is just some extra glue, it is not a driver
<leviathanch_> Turl: i'm making a prepare_enable
<leviathanch_> when I power on
<leviathanch_> the host
<leviathanch_> in set_ios
<leviathanch_> from the mmchost interface of the linux kernel
<wingrime_> libv: sdl2 can run on top gles, so may soft will be work nicely
<wingrime_> libv: cool news
<Turl> leviathanch_: so you can enable it just fine, what's the problem?
<leviathanch_> I have to disable the clock before I set a new rate?
<tm512> sdl2 has an accelerated 2D mode, that few games run on top of, because sdl2 isn't released yet
<leviathanch_> Turl: I can't disable it btw
<leviathanch_> it hangs
<leviathanch_> ;-)
<leviathanch_> or was it
<leviathanch_> oops
<leviathanch_> never mind
<leviathanch_> I'll test again
<leviathanch_> sec
<leviathanch_> nope
<leviathanch_> disable leads to a hang
<leviathanch_> ...
<wingrime_> tm512: at least openTTD
<leviathanch_> Turl: is the mod0 maybe disabling its parent as well?
<leviathanch_> Turl: this would mean it would disable the RAM clock
<leviathanch_> ...
<Turl> I'd hope the framework doesn't do that
<Turl> it has the CLK_IGNORE_UNUSED bit set, it shouldn't
fredy has quit [Excess Flood]
<Turl> looks like it doesn't take that into account in the framework :(
fredy has joined #linux-sunxi
<Turl> leviathanch_: try
<leviathanch_> Turl: know what would be way easier?
<leviathanch_> Turl: if I would just select the parent clock in the mmc driver
<leviathanch_> ;-)
<leviathanch_> but on the other hand
<leviathanch_> sooner or later
<leviathanch_> mainline has to learn that it shouldn't shut down clocks which are used for essential periphery
<Turl> some driver should claim them, then they'd remain on
<Turl> but for now we have CLK_IGNORE_UNUSED
<Turl> but the framework ignores it on the disable path
<oliv3r> tm512: how do the mmc driver and the disp driver relate?
<Turl> mturquette: ping
<Turl> oliv3r: they both make your system useful :p
<mturquette> Turl: pong
<Turl> mturquette: the framework currently ignores CLK_IGNORE_UNUSED on the clk_disable path
<Turl> mturquette: is it a bug or meant to be that way?
<tm512> id be booting off of microsd, i think id need the driver for that
<oliv3r> Turl: :p
<oliv3r> tm512: boot from nfs
<arokux2> tm512, do you also need usb or wifi?
<oliv3r> tm512: or boot with initramfs, that's what we do :)
geecko has quit [Ping timeout: 240 seconds]
<oliv3r> tm512: but yeah, it would be great if you'd started on the disp driver and KMSing it
<tm512> usb yes, maybe wifi (ath9k)
<oliv3r> there's emac for now to get you started though
<tm512> the initrd would need the mmc driver to mount the card
<oliv3r> baby steps :)
<oliv3r> tm512: nah why
<oliv3r> tm512: U-boot boots the bootloader from the SD card, u-boot then loads the kernel + initramfs and then that could mount nfs if you wanted it to
<oliv3r> or just boot via tftp and forgo sd
<tm512> also, typing this on my tablet, so im slower
<tm512> just want to boot off of sd, and mount nfs for extra storage
<oliv3r> tm512: shouldn't be a problem
<Legitsu> tacos are cool
<ojn> Hmm, A23 would be a nice speed bump compared to A20. I wonder if they have any other surprises in it.
<tm512> and ill go with whatever kernel works
<tm512> if i can basically copy the drivers over to a local mainline brach, cool
<oliv3r> ojn: A23?!
<Turl> oliv3r: what?
<oliv3r> SUPPRISE!
<Turl> ojn: how come you get to know that things before us? :)
<libv> smaller process?
<oliv3r> better memory controller?
<Turl> ojn: afaik the 1.2Ghz of the A20 is a lie, it runs @900something
<oliv3r> where's our cubie2.3
<oliv3r> Tsvetan: ^^
<Turl> oliv3r: they just add the gpu freq up to market it :)
<oliv3r> Turl: so it's ONLY a higher clockspeed?
<Legitsu> dual core ?
* Legitsu meh's
<Legitsu> wheres the 6 and 8 core chips
<oliv3r> and it's scheduled to go into mass production in October.
<oliv3r> it doesn't exist yet
<oliv3r> pff
<Turl> oliv3r: I dunno, can't read chinese news sites :P
paulk-aldrin has quit [Quit: Ex-Chat]
<oliv3r> the chip will use 50 percent less power than some (unspecified) competing chips when playing music. heh
<Turl> oliv3r: core i7? :P
<ojn> Turl: It was posted on g+
<ojn> i.e. i follow liliputing there
<oliv3r> if it now does cedarA for mp3 :p
fredy has quit [Excess Flood]
<Turl> ojn: :)
<Turl> ojn: DDR600 according to goog translate
fredy has joined #linux-sunxi
<Turl> err oliv3r ^
<tm512> oh, and I forgot to ask, the cubie will work with a modern DVI monitor, with an hdmi-dvi cable, right?
<Turl> irssi gets you two confused
<Turl> interesting, MSI shipping AW hardware
<oliv3r> tm512: HDMI is DVI + hdcp (optional) + audio; so of course it will
<oliv3r> Turl: what?!
<Turl> A20 tab
<tm512> oliv3r: I read some people had issues with it not displaying, or giving incorrect colors
<oliv3r> wires and signals are identical
<tm512> just going off of what I read
<wingrime_> oliv3r: I hope a23 not like a13
<wingrime_> oliv3r: but it looks to be cheap version of a23
<oliv3r> Turl: i'm supprised we didn't even know about this before
<oliv3r> wingrime_: i think it's just a higer clocked part
<oliv3r> it's about 300 MHz faster
<oliv3r> but I do hope it's a respin with fixed parts and a little bit faster ddr
<wingrime_> oliv3r: thats fixed a20 that run 1.2 ghz?
<Legitsu> fast ram makes a world of differnce in gpu performace
<techn_> bigger cache :)
<Legitsu> cpu speed is illrelavent in most cases
<leviathanch_> Turl: ok, it seems to be reading correctly
<leviathanch_> last problem, then we should have a functioning driver
<leviathanch_> interrupts
<leviathanch_> when it is done reading
<leviathanch_> the controller produces an interrupt
<wingrime_> Turl: are patches sended?
<wingrime_> Turl: for clocks?
<oliv3r> i'm appalled we didn't know about this :(
<oliv3r> well hipboy did say there would be a respin
<arokux2> you have expected allwinner will inform us?
<arokux2> we are nothing to them ;) just free labor
<Turl> wingrime_: not yet, I pushed draft branch if you want to take a look though
<wingrime_> Turl: realy imortant have it before any driver
<Turl> wingrime_: I know
<Turl> wingrime_: really important it works correctly too :)
<Turl> wingrime_: I found a few bugs with leviathanch_ testing the code
notmart has quit [Quit: notmart terminated!]
<wingrime_> Turl: also, about possible feed tree
<arokux2> have ever got something from AW directly?
<wingrime_> Turl: can we manipulate tree for lowpower
<Turl> wingrime_: feed?
<oliv3r> we got git history leaked
<wingrime_> Turl: clock feed tree
<Turl> wingrime_: what about it?
<arokux2> leaked isn't officially given
<wingrime_> Turl: clock source select
<wingrime_> Turl: pll select
<oliv3r> arokux2: they would never take that liability on themselves
<oliv3r> but unofficially we've gotten tons
<wingrime_> Turl: quartz feed all clocksources
<wingrime_> arokux2: docs are realy usefull
<arokux2> but they are also from third parties?
<Turl> wingrime_: you can reparent cpu to osc32k then turn off osc24M
<wingrime_> Turl: thats simple, but if I want reparent some hw to some pll , shutdown some pll for less power
<wingrime_> Turl: I talking about automatic pll select
<oliv3r> arokux2: not really
<oliv3r> but not from aw officially of course
<wingrime_> Turl: automatic pll on/off
<Turl> wingrime_: if pll not used, pll off
<Turl> wingrime_: all clocks not used -> off
<arokux2> why they cannot act officialy? they do not have bosses :)
<wingrime_> Turl: but can be pll automaticaly choosen
<oliv3r> arokux2: chinese are very carefull, they don't wanna get sued
<Turl> wingrime_: based on what criteria?
<arokux2> oliv3r, by whom?? we should sue those gpl violators in the first place
<wingrime_> arokux2: there more nice MCU like msp430 you can choose 2 quartz sources/internal/ with many combination
<arokux2> wingrime_, ?
<wingrime_> Turl: lowpower , make less pll used same time
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Soru__ has joined #linux-sunxi
<oliv3r> arokux2: ANYbody
<Turl> wingrime_: suspend?
<oliv3r> arokux2: sot hey let stuff leak and pretend its not theirs; its just chinese culture i guess
<wingrime_> Turl: no,
Soru_ has quit [Ping timeout: 245 seconds]
<wingrime_> Turl: if you have 2 device that run same speed but use 2 pll, why not reconnect it to one
<arokux2> oliv3r, ok.. I do not care so much. what I know is that there wasn't anything done by them officially for
ZetaNeta has quit [Ping timeout: 245 seconds]
<wingrime_> Turl: or simular, if device have some dividors , so better use it if there is no reason to pll
<oliv3r> engineers want to but aren't allowed by middle management
<oliv3r> they did officially allow docs to be used
<oliv3r> so hoepfully we get a23 docs soon :)
<oliv3r> Tsvetan: ^
<arokux2> we have to much docs and almost no code ;P
<wingrime_> Turl: if requested freq can't be done with diviors, automaticly use pll
<wingrime_> Turl: thats realy possible?
<wingrime_> arokux2: worser
<wingrime_> arokux2: for some hw we have no --- even blobs
<wingrime_> arokux2: no diver/no blob/no docs
<Turl> wingrime_: not all devices connected to all pll
<wingrime_> Turl: yeax
<Turl> wingrime_: sometimes frequency incompatible, can't be produced by 1 pll and two devices
<oliv3r> arokux2: so less talking more coding :)
<Turl> wingrime_: would need lot of black magic code overall
<oliv3r> arokux2: we did get a lot of code drops
<arokux2> oliv3r, no, I mean mainlining is slow
<arokux2> :)
<wingrime_> Turl: aw too lazy add multiplexers for make possible use feed tree more accuracy and make chip powerlow frendly
<wingrime_> Turl: like MSP430
<oliv3r> arokux2: code harder :)
<arokux2> I do, but I'm a noob (
akaizen has joined #linux-sunxi
<mnemoc> wow... even after removed the reverts and squashing the hotfixes stage 3.4 is 240 commits ahead of non-stage
<wingrime_> oliv3r: wow is killed
<arokux2> maybe it is better to release more frequently so that the code gets tested
<arokux2> wingrime_, killed?
<wingrime_> mnemoc: I still want usb patch be accepted
<wingrime_> arokux2: see page
<arokux2> wingrime_, what do you mean by killed?
<wingrime_> arokux2: can't contact db, too many users
<oliv3r> wingrime_: its on chinese server now
<arokux2> wigyori, works for me
<wingrime_> arokux2: slow as hell
<arokux2> well, it just contains links to our wiki
<oliv3r> wingrime_: slowish but works; they use a .cn host now
<arokux2> so why do you bother?
<wingrime_> arokux2: tons spam
<arokux2> wingrime_, yep :)
<arokux2> why could be found in a book about cubieboard? I do not understand anyway...
<Turl> wingrime_: yes, has had spam since long time
<arokux2> what*
<arokux2> how can I find which driver supports 05e3:0608 Genesys Logic, Inc. USB-2.0 4-Port HUB
<arokux2> in the mainline?
<Turl> arokux2: howto for high level dev
<arokux2> wingrime_, it's truck
<oliv3r> allwinner arm news :D
<leviathanch_> Turl: how did you determin the interrupt number?
<wingrime_> oliv3r: what a IC near VGA
<arokux2> Turl, which highlevel dev? :)
<hno> arokux2, it's a usb hub...
<arokux2> hno, yes :0
<oliv3r> leviathanch_: usermanual :)
<wingrime_> oliv3r: oh, spdif
<arokux2> hno, should it be supported out of the box?
<Turl> arokux2: people who want to toggle gpio or use led
<Turl> leviathanch_: determine interrupt for what?
popolon has quit [Quit: Quitte]
<hno> arokux2, it's supported out of the box.
<wingrime_> Turl: I have never saw conditions when it required
<wingrime_> Turl: a20 can be so hot?
<hno> USB standard defines how hubs operate. No special drivers needed for different brands.
<oliv3r> wingrime_: a10 too; specially when you OC it
<Turl> wingrime_: have not played with A20 much
<leviathanch_> Turl: for the sd/mmc 0 controller
<mnemoc> wingrime_: as I can't do more than build tests, you need to get the devs in the ML to accept it
<leviathanch_> but did it right anyway
<leviathanch_> was correct with just substracting the offset
<leviathanch_> hmm
<leviathanch_> ok
<mnemoc> wingrime_: once they ack, i take it
akaizen has quit [Read error: Connection reset by peer]
<Turl> leviathanch_: I never worked with interrupts, I don't know
<oliv3r> mnemoc: only so little dev talk on the ML
<Turl> leviathanch_: I recall seeing docs on Documentation/ about it
<wingrime_> oliv3r: please test usb fix and mail something!!
<mnemoc> wingrime_: it's not my mere whimp. it's community choice. I see nacks or disagreements, patch delayed until v2
<oliv3r> well the usermanual should have the info or the link I pasted above should be something
<oliv3r> wingrime_: lol i don't even know how to trigger the bug :p
<wingrime_> oliv3r: try select otg only
<oliv3r> wingrime_: i'll read maila nd try to reply sensibly
hramrach has quit [Ping timeout: 240 seconds]
<mnemoc> oliv3r: no need to discuss, ack mails are enough for me
<oliv3r> wingrime_: oh i remember that bug
<mnemoc> but a nack is a blocker
<mnemoc> and both wingrime_ patches have nacks
<wingrime_> mnemoc: yes, in first case, I can agree with ssvb that it can be workarounded in xorg driver
<wingrime_> mnemoc: but in second I not see any
<mnemoc> wingrime_: i can not argue because i can not test
akaizen has joined #linux-sunxi
<wingrime_> mnemoc: thats bug can be easy be triggerd
<arokux2> thanks hno
<mnemoc> wingrime_: get techn's ack on this or a v2 and it's done
<wingrime_> techn_: !!
<techn_> I'll review it
<mnemoc> in the mean time, let's merge!
<wingrime_> techn_: I agree that possible select otg but not trivial, also, it makes "device only" builds
<leviathanch_> Turl: ok
<hno> mnemoc, stage -> main btanch merge time?
<leviathanch_> Turl: oliv3r has posted me the specific part I needed
<leviathanch_> but still
<leviathanch_> the interrupt doesn't appear to be firing
hramrach has joined #linux-sunxi
<wingrime_> techn_: that patch makes select mode more easy, as it "selects" not "requires"
<techn_> wingrime_: it has many other changes
<techn_> which causes regression
<mnemoc> hno: yes. after removing a reverted patch from arokux2 and squashing a fix from Turl
<mnemoc> 240 commits
<mnemoc> unless you object
<hno> good. Have long recommended people to look at stage instead of main branch.
<wingrime_> techn_: regression?
<Turl> leviathanch_: did you enable it?
<techn_> I'll send review soon
<mnemoc> making a final free build for all defconfigs, and then the push
<leviathanch_> Turl: enable?
<mnemoc> s/free/fresh/
<leviathanch_> Turl: I enable it in the code
<leviathanch_> or
<leviathanch_> ouch
<wingrime_> techn_: only thing I have added to files --- #ifdef for make device only builds
<hno> hm... defconfigs... wonder if those have been fixed. Last time I tried there was a bit of issues there.
<leviathanch_> Turl: do I need to extract it with of?
<leviathanch_> Turl: smc_host->irq=platform_get_irq(pdev, 0);
<leviathanch_> Turl: I'm doing this
<arokux2> hno, any idea why? --->
<leviathanch_> Turl: but the IRQ is defined within device tree
<hno> arokux, not really, but I would assume there is nothing connected.
<Turl> leviathanch_: that should be enough I guess
<Turl> leviathanch_: are you assigning an irq handler then?
<leviathanch_> yes
<leviathanch_> maybe the IRQ doesn't fire because there is no data to get?
<leviathanch_> hmm
<arokux2> hno, hm.. there should be.. on-board hub
<Turl> make sure you got the irq right, check /proc/interrupts
<leviathanch_> Turl: 64: 0 GIC 64 sunxi-mmc
<leviathanch_> yes
<leviathanch_> it's correct
<leviathanch_> hmm
<wingrime_> mnemoc: 240 wow
<ssvb> wingrime_: testing the workaround now
<mnemoc> long time without merging
<wingrime_> ssvb: in driver?
<mnemoc> wanted to make a 3.4.y jump too, but will wait some days before tagging/jumping
_BJFreeman has joined #linux-sunxi
<ssvb> wingrime_: in xorg, and also the tweak to allocate scalers only when really needed
_BJFreeman is now known as BJfreeman
<wingrime_> ssvb: thanks, I now can try merge cedar driver with a13
<wingrime_> ssvb: as I can test it
<wingrime_> ssvb: also, are you tryed enable one scalar in fex?
<wingrime_> techn_: device only was not made by be
<techn_> OTG != device mode
<techn_> .. device only mode
<wingrime_> techn_: it not builds
<wingrime_> techn_: without that I made
<hno> Whole USB config in stage defconfigs seems a bit messed up, and differs between the different configs.
<techn_> if device only mode is not building it should be fixed.. not changed to other kind config
<wingrime_> techn_: ok, send patch yourself, I have no intension send v2
<techn_> wingrime_: but true.. device only mode seems wrong
<techn_> currently it seems to dependant of USB_SW_SUNXI_UDC0 .. shouldn't it be HCD?
<wingrime_> techn_: you can remake my patch,
<wingrime_> techn_: UDC is gadget
<mnemoc> To
<mnemoc> e75be34..1e8c5d7 sunxi-3.4 -> sunxi-3.4
<mnemoc> + 69181e3...18f10f5 stage/sunxi-3.4 -> stage/sunxi-3.4 (forced update)
<techn_> yes.. so I'm wondering why host only mode is dependand of gadget mode
<wingrime_> techn_: I fixed this
<techn_> ah.. I read wrong.. it is correct
<techn_> time to go sleep
<wingrime_> techn_: please fix it
<arokux2> hno, I think usb host controllers are working now
<arokux2> hno, but they fail to see something connected to them, any idea?
<techn_> wingrime_: what is wrong with it?
<mnemoc> good night. pending-but-ack patches tomorrow
<mnemoc> good night. pending-but-acked patches tomorrow
<wingrime_> techn_: #if SUN5I in #if sun4i
<arokux2> hno, I have a multimeter, can I measure smth?
<techn_> wingrime_: yes.. that whole block is inside #if 0
<hno> mnemoc, sun7i_defonfig have the otg controller disabled entirely. Is that intentional?
<techn_> wingrime_: code in comments/documentation
<wingrime_> techn_: remake my patch,
<mnemoc> hno: no, i received it like that from hansg
<mnemoc> feel free to fix
<hno> arokux2, not an expert on USB stuff.
<hno> mnemoc, default should be OTG, right?
<mnemoc> i think so
<hno> none of them defaults to OTG, sun4i and sun5i defaults to HOST_ONLY.
<hno> and sun7 defaults to none..
<mnemoc> :(
<wingrime_> hno: we need default to be in Kconfig
<mnemoc> OTG for all!
<wingrime_> mnemoc: I tryed to fix it ((
<hno> my config changes compared to sun7i_defconfig:
<wingrime_> hno: also, config normaly correctly updates if default chooses configured
<hno> wingrime_, I don't grok what you said. Can you rephrase?
<arokux2> I have better stayed with experts
<wingrime_> hno: old config normaly works if you code Kconfig right
<wingrime_> hno: with defaults
<hno> yes.
<hno> but not even using an old config.. only looking at fresh defconfigs.
<wingrime_> hno: nice, but sound still disabled by default
<wingrime_> hno: can you review
<wingrime_> oliv3r: I want notebook with 4xcortex15 from aw
<wingrime_> hno: also it need to be tested on a20
wingrime_ has quit [Ping timeout: 245 seconds]
<hno> I don't think that's the right link.. that is sunxi:disp Fully dynamic soc select
\\Mr_C\\ has quit []
<hno> night
<arokux2> hno,
<arokux2> still there?
<arokux2> 3.3V is enough for a hub?
<libv> anyone still using ubuntu 12.04 (precise pangolin) on their sunxi hw?
pirea has quit [Read error: No route to host]
<libv> heh, precise has a mesa that is older than debian stable :)
<libv> but that's only slightly
<libv> seems i have my support baseline then :)