hno changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See | | Logs at
tinti_ has joined #linux-sunxi
mahdichi has joined #linux-sunxi
mahdichi has quit [Client Quit]
BJfreeman has quit [Quit: had a good time]
uwe__ has joined #linux-sunxi
uwe_ has quit [Ping timeout: 245 seconds]
tinti_ has quit [Ping timeout: 256 seconds]
<Turl> mturquette: are you still here? :)
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
rz2k has quit []
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
pirea has quit [Quit: Leaving]
bsdfox has quit [Ping timeout: 246 seconds]
wingrime has joined #linux-sunxi
\\Mr_C\\ has quit []
deasy has quit [Quit: Quitte]
\\Mr_C\\ has joined #linux-sunxi
BJfreeman has quit [Ping timeout: 268 seconds]
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<mturquette> Turl: sorry, havne't been checking irc much today
<mturquette> Turl: but i'm always idling and logging, so just go ahead and ask your question
<Turl> mripard: np
<Turl> mripard: I've started to write a PM driver, and I require to get a hold of some clocks for it
<Turl> usually, if it were a normal driver I'd just use clocks = <...> on dt and be done with it, but pm is not a device as such
<mturquette> Turl: i think you are auto-completing mripard, but do you mean to auto-complete mturquette ??
<Turl> err yes, sorry mripard
<Turl> I got a bit used to pidgin's autocomplete and irssi's one is different :)
<mturquette> so the driver you are writing, does it have a corresponding DT node?
<Turl> mturquette: not really
BJfreeman has quit [Client Quit]
<Turl> mturquette: unless I stuff the clocks somewhere else, like on a cpu node or soc bus, but I don't think it's a very clean solution
Offshore has joined #linux-sunxi
bsdfox has joined #linux-sunxi
hipboi has joined #linux-sunxi
jinzo has quit [Ping timeout: 240 seconds]
Tsvetan has joined #linux-sunxi
<Turl> good night
<Tsvetan> and good morning from me :)
Soru has joined #linux-sunxi
<hramrach__> hello
<hramrach__> ssvb: I tried some tests with wemac
<hramrach__> finally got the iperf working properly
<hramrach__> results so far: my reference 2 weeks old 3.4 kernel on a10 is broken. current 3.4 kernel on a10 (hansg a20 branch) is broken current 3.4 kernel on a20 is broken. reference 3.3 kernel on a20 works.
<hramrach__> lpa is the same every time wemac wemac.0: eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
<hramrach__> patch does not apply - I don't have the phy_link_check function
<hramrach__> applying the patch with some manula intervention raises speed from hundreds kbit to 30 mbit
<hramrach__> the reference 3.3 kernel does 60mbit so there is room for improvement
<hramrach__> I get 3 EMAC PHY Link waiting...... so the wait would be like 1.5-2s
Soru has quit [Ping timeout: 264 seconds]
eebrah|away is now known as eebrah
atiti__ has quit [Ping timeout: 245 seconds]
<hramrach__> ssvb: on a20 manyally applying the patch gives the 60mbit again and only one EMAC PHY Linked...
<hramrach__> so there is some hw difference I guess
<hramrach__> in general the patch does improve things a lot on cubieboard
rellla has joined #linux-sunxi
jinzo has joined #linux-sunxi
<mturquette> Turl: can you share the pm driver code with me? will make it easier for me to understand.
<oliv3r> Turl: good night!
hipboi has quit [Ping timeout: 256 seconds]
n01 has joined #linux-sunxi
eebrah is now known as eebrah|away
wingrime has quit [Ping timeout: 276 seconds]
<ssvb> hramrach__: thanks for testing, kernel 3.3 handles everything related to duplex mode in a different way
\\Mr-C\\ has joined #linux-sunxi
andoma_ has joined #linux-sunxi
Liberato1 has joined #linux-sunxi
arokux1 has joined #linux-sunxi
\\Mr_C\\ has quit [Ping timeout: 240 seconds]
Liberator has quit [Ping timeout: 240 seconds]
andoma has quit [Ping timeout: 240 seconds]
arokux2 has quit [Ping timeout: 240 seconds]
<ssvb> hramrach__: btw, looks like I found some way to workaround that mali dri2 window resize bug -
<ssvb> hramrach__: which means that unless some other problem shows up, we can eventually have zero copy buffers swapping and vsync for mali x11
FR^2 has joined #linux-sunxi
hipboi has joined #linux-sunxi
rings_IIV has joined #linux-sunxi
notmart has joined #linux-sunxi
bsdfox has quit [Ping timeout: 256 seconds]
<hramrach__> ssvb: 3.4 kernel with the patch does about same in iperf as the 3.3 kernel on a20
<hramrach__> but unpatched it is half duplex
Black_Horseman has quit [Ping timeout: 256 seconds]
<arokux1> hm.. how does EOMA-68 correlate with such standards as COM Express, ESMexpress etc? (see
uwe__ is now known as uwe_
<arokux1> what happened to BDD Group A10 SoM? no news since March.
\\Mr-C\\ has quit []
tinti_ has joined #linux-sunxi
<arokux1> i've answered the question about COM_Express myself
<oliv3r> lol ok
tzafrir has joined #linux-sunxi
rellla has quit [Ping timeout: 246 seconds]
<arokux1> why is this so expensive?
<arokux1> US$89
<oliv3r> nichee market
<arokux1> why cannot I find a cheap router like board
<arokux1> with wifi module and 2xethernet
atiti__ has joined #linux-sunxi
<oliv3r> 2x ethernet is the tricky bit; there's really realyl cheap router with 1x wifi, 1x ethernet
<oliv3r> and 1x usb
<arokux1> well, at least 2xethernet
<arokux1> could be more
<arokux1> oliv3r, like this, but cheaper:
<oliv3r> 14 USD
<oliv3r> they run openwrt like a champ
<oliv3r> WR703N is pretty awesome imo
<arokux1> oliv3r, I somehow do not like the idea of messing around with openwrt
<oliv3r> openwrt is awesome
<oliv3r> all my routers run openwrt
<arokux1> full blown distro would be much better
<oliv3r> and I wouldn't want it any other way
<oliv3r> i actually even have some virtual machines running as webservers, openwrt
<oliv3r> 15mb for the OS is pretty cool, with aapche, mysql etc
<arokux1> oliv3r, wouldn't you want you routers to run debian? :)
<oliv3r> no
<oliv3r> openwrt is tuned and optimized for routing of networking packets
<oliv3r> though I wouldn't be supprised this would run emdebian (if there's enough space) just fine
<oliv3r> and if using it only as a router, why run a full blown distro, wasting flash space :p
<oliv3r> not to mention there's not even a nice GUI :)
<arokux1> oliv3r, where?
rellla has joined #linux-sunxi
rz2k has joined #linux-sunxi
<oliv3r> arokux1: in debian; openwrt has a nice gui (to configure your wifi/firewall
<arokux1> oliv3r, this is not so important for me
<arokux1> oliv3r, i just find it amazing that the router like hardware could be actually only bought most cheaply as router!
<arokux1> (dlink, tplink etc)
<arokux1> then one need to smash the stock software and flash it with openwrt
<oliv3r> yeah :)
<oliv3r> i wonder how usable a tp-link is as debian
<arokux1> oliv3r, why there is no JUST hardware, without software...?
<arokux1> it should be cheaper.. or?
<oliv3r> you'd think
<oliv3r> but remember, consumers (users) want something they buy, plugin and it has to work
<oliv3r> then most of these manufacturers buy the whole design, chiP + software
<mripard> arokux1: no one would buy it except a few geeks, production would be very low, the hardware would be more expensive.
<oliv3r> so it's 'done' mostly as they buy it
<oliv3r> mripard: Hi
<arokux1> mripard, and could tp link just offer hacker friendly version of their devices without their software with reduced price? :)
<mripard> arokux1: why ? they have no interest in doing so
<arokux1> mripard, because their software sucks and i do not want to pay for it :((
<oliv3r> how much is their software worth do you think?
<oliv3r> They budget maybe 2 cents per router for the software?
<oliv3r> which is why its so totally crap most of the time
<mripard> arokux1: yeah, but the software cost is almost null, so you wouldn't benefit from it
<mripard> while having to handle this in production is an extra burden. and an additionnal cost
<arokux1> oliv3r, and then they prohibit others to develop something else
<arokux1> WHY?
<oliv3r> they don't care
<oliv3r> that's why
<oliv3r> they buy the chip, they get a "free" sdk with it from the chip manufacturer
<oliv3r> they change the options/settings so that the firmware supports their hardware configuration combination
<oliv3r> (e.g. wifi chip A, ram chip B, flash chip C)
<arokux1> oliv3r, i'm not sure, but my impression was that they do not like see people develop openwrt etc
<oliv3r> burn that image to millions of flash chips at the factory, and sell their device
<oliv3r> if they have to sell unflashed versions, they have to a) badget them seperatly (costs money) b) make sure only the right people buy it etc etc
<oliv3r> arokux1: of course not, you are doing something with their prodcut that might break it, or brick it and then you come back crying for warranty
<oliv3r> which costs them money
<arokux1> oliv3r, and it's not easy to understand what happened to the product?
<oliv3r> investigation == time
<oliv3r> time == money
<arokux1> oliv3r, they need to investigate anyway
<oliv3r> why?
<oliv3r> you are doing something unsupported, you should go way, bye bye and good luck
<arokux1> yes, but if somebody brings the device and says it's not working they check if it was used properly..
<arokux1> if still sealed and stuff
<arokux1> the could check if the firmware is still in place, computing some checksum..
<rz2k> no one does that for a long time
<mripard> arokux1: it's also a matter of cost and statistics
<mripard> if the router costs 50$
<mripard> that checking for the firmware costs 1$
<rz2k> logitech for example just sends you replacement without many questions asked
<mripard> and that only 1% of the returned router is actually returned because you flashed a broken firmware
<Turl> mturquette: as long as you don't get scared by ugly code, sure :P
<mripard> the vendor has to pay 100$, while it gained 50$
<mripard> so in the end, he lost 50$
<mripard> so why doing the check in the first place?
<rz2k> seagate also sends you replacement hdd's after S/N picture sent and info about whats wrong
<oliv3r> Turl: good morning
<Turl> morning oliv3r
<Turl> mturquette:
<arokux1> ok, so the 2% that are not in mainstream are just screwed up..
<oliv3r> you are drifting from your own original point. Why is it not cheaper without software, very small market, software cost is nearly zero, if not zero
<oliv3r> Why don't they want you to run anything but their firmware, because you messing with it and sending it back costs them money. You not touching it but only 'using' it in the warranty period is good for them
<arokux1> oliv3r, hm.. could their be an option to refuse to have warranty from the very beginning
<oliv3r> nope
<mripard> options cost money.
<arokux1> oliv3r, if I refuse from warranty they send me a magic key how to load new software
<oliv3r> jinx mripard
<oliv3r> arokux1: but how is that an issue
<oliv3r> 99% of the routers can be re-installed with custom roms anyway
<oliv3r> they don't really protect it most of the time
<mripard> ah, so your point isn't about hardware without software, it's about you being able to flash a new firmware
<oliv3r> all you need is a method to flash the firmware, either via serial port or via spi if your really desperate
<mripard> or JTAG :)
<oliv3r> or jtag :)
<oliv3r> besdies, if it's really a big issue
<oliv3r> they most of the time use u-boot and/or linux; which you can get nearly all ifno for, unless they are GPL infringing scumbags
<oliv3r> anyway, your point about refusing warranty if they send you amgic key
<oliv3r> ask how micrsoft will like that
<oliv3r> i want this windows surface thing, but i want to run linux/android on it, give me the uefi key, i hereby refuse warranty
<oliv3r> i bet microsoft will give you the key without questions asked! :)
<arokux1> oliv3r, they still sell hardware :P
<oliv3r> seriously though, there's actually NL law (european too?) about this kind of behaviour from manucaturers. ROoting a phone, flashing a custom rom on it, is legally allowed and they (samsung etc) are not allowed to deny you warranty
<oliv3r> arokux1: lol in microsofts case, they don't give a rats ass about the hardware sale; they want world domination
<oliv3r> so if you run android on their tablet, that makes them very unhappy
Brod_ is now known as Brodomir
Brodomir has quit [Quit: - Chat comfortably. Anywhere.]
Brodomir has joined #linux-sunxi
<arokux1> after much of googling the cheapest hardware capable of running a stock distro with 2xEthernet is Utilite ($99) and DreamPlug($159).
<oliv3r> dual ethernet is pretty 'big' requirement
<oliv3r> i would imagine any openwrt-capable router with USB port should be able to boot a full distro via USB
<arokux1> but that router will cost more
<arokux1> i think
<oliv3r> wellt he TP link is only single port, but has usb port :)
<oliv3r> most accesspoints only have 1 ethernet port
<Turl> oliv3r: WR1043ND probably costs half that
<oliv3r> they usually have an internal 'switch'
<oliv3r> Turl: which one ist he wr1043ND?
<Turl> 4xGbE and wan, as well as wifi N :p
<oliv3r> but that's not dual ethernet, it's single + switch
<Turl> well, you get 5 jacks don't you? :)
<Turl> it can do VLANing
<arokux1> Turl, WR1043ND yes, but then maybe TP-Link TL-WDR3600 because of better hardware
<oliv3r> the wr702n is even quarter of that :)
<arokux1> oliv3r, it's not only about the router, it should be able to supply decent hard disk access speed
<oliv3r> well your mostly limited by USB there
<oliv3r> or NFS
<oliv3r> routers with sata are very rare
<oliv3r> your better of getting a NAS with dual LAN ports and do wifi via usb
<rellla> if someone can tell me his opinion if color space can be the blue-issue of xbmc?
<arokux1> usb is enough, but i've heard those weak routers will give you 2MB/s or so
naobsd has joined #linux-sunxi
<Turl> arokux1: what about a sunxi device with a usb ethernet thing?
<arokux1> Turl, yes, this is what I will use.
<arokux1> i just couldn't believe i cannot find some nice cheap board with 2 ethernet ports..
<Turl> arokux1: I've asked around, I'll let you know if I find sth :p
<arokux1> Turl, thanks, not that it's important, just out of interest
<Brodomir> Quick Question, If i have a 32GB SD HC is there an advantage to installing debian to the NAND or just run everything on the SD?
bsdfox has joined #linux-sunxi
bsdfox has quit [Changing host]
bsdfox has joined #linux-sunxi
<oliv3r> hno: i think i properly split the patch now; if i overlook something else, please do say so :)
<oliv3r> not only is it quite expensive, to me a nother 'killer' and discrimination is the "free shipping to US and UK, 30 USD to everybody else"
<oliv3r> UK is a stonethrow away, but I need to pay 30? forget that
<arokux1> UK that's maybe because Canonical is UK based
<arokux1> oliv3r, but still, it's very cheap now. S5 will be probably more expensive :)
WarheadsSE has quit [Quit: WeeChat 0.4.0]
WarheadsSE has joined #linux-sunxi
<oliv3r> South Africa based*
<oliv3r> 625 USD is a lot of money for me :)
<arokux1> i'm talking about the price of the hardware
<arokux1> still, i do not believe they will succeed
<oliv3r> 32 million is a huge amount
Undertasker has joined #linux-sunxi
Undertasker has left #linux-sunxi [#linux-sunxi]
<arokux1> oliv3r, it's not the amount but the price
<oliv3r> price is okay i suppose for a super highend smart phone
<arokux1> now yes
<oliv3r> s4/s5 costs more, you are right
<oliv3r> but I don't buy phones that cost that much :)
<arokux1> you could fund 100M if the price of such phone would be $500
<oliv3r> i buy older models when they are cheap
<arokux1> oliv3r, me too
<oliv3r> i'm getting an s2 next week or so
<oliv3r> i currently have an htc hero :p
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<mripard> oliv3r: Canonical is based in the Isle of Man, UK
<mripard> Mark Shuttleworth is south african, but that's all
<arokux1> yes
<oliv3r> oh really; I did not know that
<oliv3r> i knew they had 'an office' there
<arokux1> :)
<mripard> oliv3r: i'm not sure they have a real office there
<mripard> the reason fits in two words
<mripard> tax haven
<oliv3r> but the whole ubuntu thing is being brought as a south african thing
<mripard> nah, the name is, that's all.
<mripard> the real HQ are in London iirc
<oliv3r> oh really; i'm dissapointed now
deasy has joined #linux-sunxi
<mripard> why? because they chose an african world as their brand?
<mripard> you should be disappointed quite often :)
<oliv3r> i am :(
<oliv3r> the world is dissapointing :(
Undertasker has joined #linux-sunxi
Undertasker has left #linux-sunxi [#linux-sunxi]
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest24794
_BJFreeman is now known as BJfreeman
Guest24794 has quit [Ping timeout: 268 seconds]
\\Mr_C\\ has joined #linux-sunxi
BJfreeman has quit [Quit: had a good time]
Undertasker has joined #linux-sunxi
Undertasker has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
Undertasker has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
deasy has quit [Quit: Quitte]
Undertasker has left #linux-sunxi [#linux-sunxi]
<slapin_nb> hi, all!
<slapin_nb> hno: are you here?
<slapin_nb> hno: have you got A20 version of cubieboard?
<slapin_nb> mripard: hi, are you here?
<slapin_nb> any news?
<slapin_nb> Aa!
<mripard> slapin_nb: hi
<slapin_nb> mripard: any news?
<slapin_nb> mripard: have you got A20 version of Cubieboard?
<slapin_nb> mripard: I'm a bit confused with bundled USB->uart converter, but anyway
<mripard> yeah, I got one
<mripard> any news about what?
<rz2k> everyone who attended to cubieboards developer program got one
<slapin_nb> <- found one, I think it is better to be easier to find...
<slapin_nb> mripard: I want to work on mainline a bit, what are current status and roadmap?
<mripard> well, we have almost complete A31 support
<mripard> the only bit missing is SMP
<mripard> A20 is booting
<mripard> I'm currently working on the GPIOs, it's almost done
<mripard> the next step will be the clocks
<mripard> Turl worked on the older SoCs clocks
<mripard> and for the roadmap, we don't have any roadmap
<mripard> if people work on stuff, we go forward, if not, well, nothing happens
<slapin_nb> mripard: I just don't want to do other people are already doing
<slapin_nb> mripard: what is status of NAND controller support?
<slapin_nb> mripard: and what are other delaying/annoying tasks which stop good things from happenning?
<rz2k> slapin_nb: nand is not happening until dma is ready
<mripard> yeah, DMA is the biggest showstoper these days
<rz2k> slapin_nb: please read
<slapin_nb> rz2k: ok, will look into this stuff, thanks!
ykchavan has joined #linux-sunxi
<rz2k> all the patches is at mripard's github, if you want to check out ones that are not in mainline right now
<rz2k> s/is/are/
<slapin_nb> rz2k: URL?
<mripard> slapin_nb: if you want something you can start working on right now and is quite easy, I guess you can start with the PWM driver
<mripard> oliv3r is supposed to work on it, but I'm not sure if he has actually started or not
<mripard> so you can always ask him
<rz2k> slapin_nb: the obvious one?
<rz2k> mripard: any news from mdp ?
<slapin_nb> rz2k: I am not big beleiver in obvious things. ?
<slapin_nb> mripard: thanks!
<mripard> rz2k: yeah, he told me last week that we should expect the SPI patches in a couple of weeks
<mripard> and he will use the SPI to bring up DMA after that
<rz2k> awesome
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 256 seconds]
rellla has quit [Quit: Nettalk6 -]
deasy has joined #linux-sunxi
<Turl> slapin_nb: a driver for buttons on boards would be nice too :)
<mripard> oh, yeah, or the LRADC driver
<mripard> Turl: great idea :)
<mripard> IIO isn't probably the easiest framework to start with
<mripard> though
atiti__ has quit [Ping timeout: 256 seconds]
<slapin_nb> mripard: I'm quite fluent with kernel, don't worry.
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<mripard> slapin_nb: ok, great then :)
<slapin_nb> IIRC LRADC is these damn buttons with resistors attached to ADC, which is intended to drive resistive touch screen?
<slapin_nb> on my last tablet I have single button attached this way...
<mripard> the touchscreen is handled by a separate IP iirc
<mripard> LRADC is mostly used for buttons yes
<slapin_nb> mripard: the capasitive touch is separate chip, not inside processor.
<mripard> ah, capacitive, probably
<mripard> but you were mentionning resistive
<slapin_nb> mripard: LRADC was intended to drive resistive touch IIRC. I currently make device which will need resistive touch and can't find driver for A13 so I could use LRADC as touch screen interface :(
<slapin_nb> resistive touch is cool for stylus and drawing, capacitive is only for fingers
<slapin_nb> and people will need to draw sketches
<mripard> slapin_nb: I'm not sure it has what's needed, like touch-down interrupts and so on
<mripard> oh, it does
<mripard> ok. nevermind.
arokux has joined #linux-sunxi
<slapin_nb> mripard: more than that it can do magic 'multitouch'-like things
<oliv3r> slapin_nb: omg welcome back :)
n01 has quit [Ping timeout: 264 seconds]
FR^2 has quit [Quit: Connection reset by peer]
<hno> slapin_nb, I have. But not with me at the moment.
hipboi_ has quit [Quit: Leaving]
<slapin_nb> hno: hi! I've resolved issues with cubie2
<hno> slapin_nb?
<hno> Ah.
<slapin_nb> hno: the USB serial thingie
<slapin_nb> hno: which is current wip branch with u-boot? I want to resume NAND work
<slapin_nb> mripard: ah, I'e forgot something - should I submit patches to linux-sunxi@ for you to pick up?
<slapin_nb> hno:
<mripard> slapin_nb: nope, send it to the usual kernel mailing list
<mripard> so whatever show up when you use get_maintainers
<mripard> and put me either in Cc or To
<slapin_nb> mripard: ok, thanks!
<mripard> slapin_nb: and usually, we put linux-sunxi in Cc
<mripard> except when we forget :)
<mripard> (like for the A31 patchset...)
<hno> slapin_nb, for a20 the current branch is wip/a20
<hno> mostly ready for merging into sunxi-current, just need to split patches a bit.
<oliv3r> slapin_nb: oh good; rz2k was working hard on nand and got some working shit, but he's ona break right now; $job got in the way
TheSeven has quit [Read error: Connection reset by peer]
TheSeven has joined #linux-sunxi
egbert has quit [Ping timeout: 268 seconds]
egbert has joined #linux-sunxi
<arokux> hno, can I force reboot by some button press on mele?
<Turl> arokux: long press power button, after it shuts down, turn on again
<arokux> Turl, thanks, it worked.
<oliv3r> slapin_nb: there's a resistive touch screen controler inside the SoC; you can also use it as an ADC otherwise
<oliv3r> slapin_nb: but mtd work would be completly awesome :)
<oliv3r> since we lack storage support in mainline atm; and yeah i have some skeleton pwm thing, but i was trying to push sid driver, and working on a31 u-boot which is a bitch i tell you
<oliv3r> mripard: wish i could push some of these changes somewhere for others to look at :S
<oliv3r> mripard: i think you did cc the a31 patchset; i saw it pop by
<oliv3r> hno: check my github; i've split the patch, unless you hint at something else i overlooked
<oliv3r> arokux: the power button is usually (almost always) connected to the AXP. A long push actually shuts down the AXP, and hence everything else
<oliv3r> and everybody is gone again :(
<slapin_nb> oliv3r: tough luck if you have additional PMICs in parallel to AXP (instead of being controlled by it) :)
* slapin_nb is afk for 2 hours
<oliv3r> source > images
<oliv3r> slapin_nb: yeah, i don't think we quite know how to reset one of those PMIC-less
<oliv3r> like A13 hdmi sticks
<oliv3r> not that you want to power them off
<Turl> oliv3r: source is pushing now :)
<oliv3r> that fast?
<oliv3r> that's new
<Turl> oliv3r: have a look at the funny announcement :)!topic/android-building/sDovKMKTAiQ
<oliv3r> google airlines; lol
<oliv3r> bah
<oliv3r> requires new drivers
<arokux> is it normal that after some time uboot crashes or the like? :(
<oliv3r> arokux: erm, kinda unlikly
<Turl> oliv3r: I don't expect many issues on that front
<arokux> how often have you used emac in uboot in mele (a1000)
<Turl> it always does :p
<oliv3r> Turl: why not
<Turl> because ^
<oliv3r> arokux: i don't have a mele
<oliv3r> Customs forms are available at and need to be filled before starting a device build, even if this is not your first trip to Jelly Bean.
<Turl> oliv3r: OpenGL ES 3
<Turl> wonder what libv has to say re. it :P
<oliv3r> well with regards to the S2, do you know if it does GL ES3?
<oliv3r> paul-desktop ^
<Turl> oliv3r: not unless a driver upgrade at least
<Turl> but no OGLES3 is not going to kill you
<oliv3r> good :)
<Turl> oliv3r: consider 99.9% of the market is OGLES2 atm :P
<oliv3r> Turl: libv was a little down yesterday, about to abandom lima n stuffs :( which made me sad and almost cry a little
<Turl> oliv3r: :/
<arokux> oliv3r, do you still use emac from uboot-current?
<oliv3r> arokux: what do you mean?
<arokux> like for network boot from uboot
<oliv3r> arokux: first off, if you boot u-boot, u-boot is active 2 seconds before handing over to the kernel, at which point, u-boot seizes to exist
<oliv3r> arokux: turl does heavy network booting, the PHY on mine is busted
<arokux> no, the kernel isn't loaded yet, I endup in uboot shell
<oliv3r> and you can't boot the kernel from u-boot?
<oliv3r> do you have uart connected?
<arokux> i can i.e. going to (meaning it is not a problem), but i thought network boot could be more fast for developing and booting the kernel each time, since sd card need not be rewritten
<oliv3r> arokux: that is true
<oliv3r> you setup a nfs server (for root fs if you need it)
<oliv3r> a tftp server
<oliv3r> then you make a boot.scr or whatever and make u-boot load the kernel via tftp
<oliv3r> you still need an SD card with u-boot-spl and u-boot on it (and u-boot.scr/env)
<oliv3r> Turl: is your expert :p
<arokux> oliv3r, yep, but uboot won't be rewritten.
<libv> mali-400 will never be opengles3 compatible, it lacks some features for that iirc
<arokux> oliv3r, question: where can i load kernel? what is free in memory space?
<oliv3r> arokux: 0x48000000
<oliv3r> 0x43000000 for script.bin
\\Mr_C\\ has quit []
<arokux> i'm going to load mainline kernel, so no script.bin
<arokux> after which 0x.. there this an empty and unused ground? :)
<oliv3r> yeah
<oliv3r> 0x43000000 for devicetree.dtb
<arokux> where is this address coming from?
<oliv3r> your keyboard :)
<arokux> oliv3r, here its 0x43..
<oliv3r> it doesn't matter
<oliv3r> memory address space starts at 0x40000000
<oliv3r> before are for registers
<arokux> ok, that is what I wanted to know :)
<arokux> its accessible up to 0xFFFFFFFF? :)
<oliv3r> you can load your kernel at 0x40000000 if you want to
<oliv3r> if you have that much ram, sure
<oliv3r> 2g is our amx address space :)
<oliv3r> but i think the a10 ram controller pukes at anything over 1g
<arokux> amx? :)
<oliv3r> ?
<arokux> <oliv3r> 2g is our amx address space :)
<mnemoc> max
<mnemoc> typos happen often on irc ;-)
<arokux> but the kernel will be extracted starting at LOADADDR=0x40008000
<arokux> so one need to be careful so that the space taken by the loaded kernel (with fatload) does not overlap with space after 0x40008000
<arokux> am i right?
<oliv3r> mnemoc: yeah especially with my add typing :p
<oliv3r> mnemoc: also, HI!
<oliv3r> correct; hence I load my kernel at 0x48000000 :)
<oliv3r> it looks like 0x40008000 and is far away from it :)
<arokux> yes, 16*8 MB far from it.
<arokux> (roughly)
<mnemoc> oliv3r: hi :)
<oliv3r> backreading mail still?
<arokux> can anybody suggest uart2usb cable to buy which doesn't let the power to flow into the board? (without a diode trick)
<oliv3r> no
<oliv3r> the board is broken, not the cable
\\Mr_C\\ has joined #linux-sunxi
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
<mnemoc> oliv3r: haven't started yet :| ... busy doing paperwork to be employable in .de, trying to get some remote $work$ done, and trying to find out why the new server doesn't want to boot :|
<mnemoc> oliv3r: is there any ready-to-pull branch I should bring in blindly?
<oliv3r> is it a virtual container again?
<oliv3r> mnemoc: hansg's
<oliv3r> mnemoc: but we have 2 regressions
<oliv3r> actually, we have a patch that brings out a pretty big bug we have had all this time
<mnemoc> oliv3r: haven't reached the lxc point yet, just want ubuntu booting on raid10/ext4 for now
<oliv3r> so imo, yes; merge hansg's, there needs to come a serious solution for the lan auto detect buga nyway
<oliv3r> is it a full real hardware server?
<mnemoc> yes
<oliv3r> do you have some kvm remote access?
<oliv3r> i had a real hardware box with OVH for a year (to test it)
<mnemoc> yes, real hw. i7 980, 24GB, 4x1.5TB, 20TB/M
<oliv3r> and they had a web interface, where you could a) load some emergency iso that always worked, and b) kvm
<oliv3r> showoff :p
<mnemoc> i'll try using hetzner's kvm/vnc based recovery thing now
<Turl> mnemoc: need a hand with anything? :)
<oliv3r> mnemoc: atleast you can see WHY it's not booting
<mnemoc> Turl: let me boot it first :p
<mnemoc> and f* btrfs
<Turl> mnemoc: :)
<oliv3r> btrfs is awesome; just not production ready yet imo
<Turl> oliv3r: zfs ftw ;)
<mnemoc> the host will be just a boring ext4. for the vservers we can experiment later
<oliv3r> Turl: no zfs for linux :( (not really anyway, stupid cddl)
<Turl> oliv3r: huh? I'm using zfs on linux right now just fine :)
<Turl> ZoL is great ;)
<arokux> Turl, which initramfs do you use here?
<Turl> oliv3r: s/cddl/lawyers/
<libv> mnemoc: wherabouts in .de?
<arokux> (this page was edited by you :) )
<Turl> arokux: buildroot
<Turl> oliv3r: did you get any news from greg yet?
<mnemoc> libv: yes, moved to berlin. need to work with flesh people
<oliv3r> Turl: nope; he said 'give me a few days'
<libv> mnemoc: nice
<oliv3r> Turl: how are you using zfs; other then zfs-fuse
<arokux> Turl, and you boot to initramfs only? I've used some miniroot from hno, and get kernel panic becase VFS: Cannot open root device "VFS: Cannot open root device ... Please append a correct "root=" boot option..."
<oliv3r> arokux: yes, you must boot either via initramfs
<oliv3r> or nfs root
<oliv3r> there's no storage possibilties yet
<arokux> oliv3r, yes, I've built initramfs into kernel
<arokux> wiki page says it should boot to shell prompt...
<oliv3r> where did you get that initramfs from? :)
<arokux> thought it would be the easiest :)
<oliv3r> arokux: i use for testing purpouses
<arokux> oliv3r, will try it out
<libv> mnemoc: unless my gf manages to convince me again, i doubt i will be pass through there any time soon :)
<libv> mnemoc: linuxtag was late may, and that is pretty much the only time i go to berlin
<oliv3r> sill need to manage to go to fosdem 2014 :)
<libv> fosdem is a must, linuxtag is not really worth it anymore
<mnemoc> ccc?
<oliv3r> i know libv but need someone to let me go first :)
<Turl> oliv3r: ZoL (ZFS on Linux), it's in-kernel ZFS :)
<libv> froscon has supposedly grown
<oliv3r> Turl: illegal patches!
<libv> i was there for the initial 2006 event, and although it was great, i never managed to go back
<oliv3r> froscon?
<libv> ccc was in hamburg last year
<oliv3r> french opensource con?
<oliv3r> chaos computer club?
<libv> sankt augustin, near bonn
<libv> ccc seems much broader than open source software
<libv> never visited it
<Turl> oliv3r: it's an out of kernel module :) and it's free software, so whatever
<Turl> oliv3r:
<arokux> why LOADADDR istn't part of the dev. tree, having it as a parameter still results in one kernel per board..
<oliv3r> it's part of the BSP
<oliv3r> but only my local branch
<oliv3r> i should push it actually
<Turl> arokux: LOADADDR is just used for "mkimage", ie, the uboot wrapping
<arokux> ah, so it's not inside of kernel, ok
* mnemoc has never been on any event of the kind
<oliv3r> meeting flesh
<oliv3r> you shoudl come this year
<oliv3r> well 2014
<mnemoc> once in cebit and once in a car wash expo in atlanta. both.... not for me
<oliv3r> fosdem is a great way to socilazie
<mnemoc> from berlin it's cheaper to go than from coruña. and finally been an employee i'll have right for vacations... so.... it will be possible to go an see the next big news about lima!
<Turl> mnemoc: they make car wash expos? :P
<mnemoc> Turl: yeah, several :p
<mnemoc> even one in las vegas
<mnemoc> atlantic city, and other sin cities
<Turl> I take it the odds are pretty much inexistant, but anyone coming to .ar for SASE next month? :P
<techn_> mripard: there is some squash errors in your a31 patch set
<Turl> techn_: I pointed one out :p
<libv> mnemoc: :)
<techn_> Turl: 6/10 and 5/10? :)
<Turl> techn_: probably, it was the makefile entry one patch before the actual dts
<techn_> Turl: same
<arokux> after loading the kernel and dtb into memory with fatload, I boot them with bootm 0x48000000 - 0x49000000, Then nothing comes after "Starting kernel ..."
<arokux> this is with builtin ramfs
<Turl> arokux: enable DEBUG_LL, EARLY_PRINTK and the sunxi option on debug_ll
<Turl> then pass earlyprintk on the boot arguments
<arokux> if it is not built in i see kernel log with "VFS: Cannot open root device"
<arokux> Turl, I think it is on
<oliv3r> arokux: aren't you overwriting your kernelw it your dtb?
<oliv3r> tot est, you can use one of my testing kernels
<Turl> arokux: also, that's like 16M apart, maybe the image is just too fat or something :p
<oliv3r> put your dtb before the kernel, 0x47000000 is safe
rellla has joined #linux-sunxi
atiti__ has joined #linux-sunxi
<arokux> your kernel worked, I haven't toched memory layout
<arokux> oliv3r, where from is this kernel?
<oliv3r> 3.10
<oliv3r> check my github
<oliv3r> but it's likly a .config issue :)
<arokux> but the difference is only initramfs......
<oliv3r> you said 'my kernel worksl'
Tartarus_ has joined #linux-sunxi
Tartarus_ has quit [Remote host closed the connection]
<arokux> oliv3r, you do not have 3.10 on your github... its 3.4 I think
<oliv3r> check wip/sunxi-security-id
<arokux> oliv3r, ok, how do you build it? ARCH=arm make multi_v7_defconfig then adding initramfs using menuconfig and make ...?
<arokux> (I have 3.11-rc1, btw)
<mnemoc> Turl: good news, the server was booting. just that my /etc/network/interfaces is f*ed up :p
<arokux> from here git://
<oliv3r> arokux: i use the sunxi-bsp :)
<oliv3r> oh yeah, i may have pulled that
<Turl> mnemoc: yay :)
<hramrach__> hello
<oliv3r> arokux: write patches against linus's dev tree
forcev has joined #linux-sunxi
<arokux> first, I need to boot kernel, then pathes...
<oliv3r> enable kernel debugging, and earlyprintk
<arokux> earlyprintk is on
FunkyPenguin has quit [Ping timeout: 264 seconds]
<oliv3r> then you should see output ;)
<arokux> oliv3r, I see it if initramfs is not included
<oliv3r> kernel boots either way, with or without
<oliv3r> and should output stuff
<arokux> i see, but with it does not.
<oliv3r> but without initramfs, you'll get root missing panic :)
<arokux> yep
<oliv3r> i extracted the cpio to /usr/src/initramfs
<oliv3r> because i wanted to modify it a tiny bit
<oliv3r> and i include that dir in my kernel build
<arokux> hm.. i include cpio, I've read it's also possible
<oliv3r> yep
<oliv3r> you include either a file, or ad irectory
<oliv3r> the kernel is smart enough to figure out what you've included
<Turl> arokux: are you passing 'earlyprintk' on kernel params?
<Turl> oliv3r: I include cpio, it works, it's not that
<arokux> Turl, yep, this is in dts
<Turl> arokux: yeah but uboot overrides it
<arokux> Turl, but it works if without initramfs..
<arokux> Turl, do you use kernel from mripard ?
<oliv3r> i don't even pass earlyprintk as para :)
<oliv3r> arokux: you can even use plain vanilla linus kernel
<oliv3r> there's support for this stuff tehrer allready
<oliv3r> your best bet imo
<arokux> so nobody of you has used mripard 's kernel?
<oliv3r> probably mripard :p
<oliv3r> actually mripard pushed my tree to github because i broke mine
<oliv3r> so technically yes I am
<arokux> oliv3r, not quite, because yours was 3.10
<arokux> i have 3..1
<arokux> 3.11
<oliv3r> ah yes, i haven't pushed my latest stuff to github
<oliv3r> that one is old :)
<arokux> anyway, you have adjusted sunxi-bsp, so it uses your kernel?
<arokux> oliv3r, because by default it will use some ancient stuff
<oliv3r> yep
<arokux> Turl, what kernel do you use?
<oliv3r> we all use our own
<oliv3r> based on linus's
<arokux> :)
<arokux> ok
<oliv3r> since we write patches against linus's
<oliv3r> well turl, mripard and I
<oliv3r> for mainline stuff that is
<oliv3r> 3.4 we use stage
<arokux> on wiki it was advised to use mripard , however he wrote that page......
utente_ has quit [Remote host closed the connection]
<oliv3r> well that is a safe choice
<arokux> i'm cloning 3.10 vanilla now, and will see
<oliv3r> 3.11-r2 :p
<oliv3r> though 3.10 has a10 stuff
<Turl> arokux: I build my own, as oliv3r says
<arokux> Turl, how, using sunxi-bsp?
<Turl> arokux: no, using "make uImage dtbs" :p
<arokux> and adding path to the cpio in the menuconfig?
<Turl> yes
notmart has quit [Quit: notmart terminated!]
<arokux> Turl, I specify full path there, can it be a problem?
<Turl> nah, I do too
<Turl> I'd start by enabling debug_ll with sunxi uart0, early printk, and passing earlyprintk from uboot
utente has joined #linux-sunxi
MadSpark has quit [Remote host closed the connection]
<utente> rellla, ssvb i 4got who of you i tal to me about my problem with audio issue in xbmc. anyway finally i worked on it, and the solution on my sistem is: AUDIO STREAM = DPLII_Q0.4 (2/2) it came out english language from hdmi.. Curiouslym audio autput is set to "analog" but the audio cames out from hdmi. also AUDIO STREMA set to AC3 5.1 vorks (in my own language)
<rellla> utente: fine
<utente> aniway sometime came out "squared wite area" all on monitor, but it happens rarely and for few seconds.
<arokux> Turl, got output: /run/media/roman/8032-2820/uImage
<utente> video reporduction (is a 480 mdkv) is little slow, i try tu push up the clock.
<arokux> Turl, sorry, this one: Error: unrecognized/unsupported machine ID (r1 = 0x00001008).
<Turl> arokux: is it a 3.11 build? I've seen that issue too, need to ask mripard about it
<utente> mmm now xbmx stopped and monitor goes int black screen :/
<Turl> arokux: go to menuconfig, look for the machine selection, and untick everything except sunxi
<utente> but in ssh from remotre cp responds.
<Turl> then rebuild, that worked for me
<arokux> Turl, yes.
<arokux> Turl, I wonder how the presence of initramfs influences it!!!
<Turl> I dunno :/
<utente> using ssh i killed xbmc and restart lightdm to have back the desktop, but now colors are falsed.
<utente> hot to restart the whole X server?
<arokux> utente, pkill X
<arokux> startx
<arokux> hope that works
<utente> "pkill X" worsk, and it restart by itself, but then colors are still false: white-->black and no login window appears
<utente> it seems xbmc need more work before to be totally usable :/
<rellla> utente: there is some color space issue imo which must be corrected in xbmc
<arokux> Turl, hm.. where is u-boot in memory?
<utente> maibe i must remove X server and user xbmc from CLI?
<rellla> thats why pink, blue and false color in X
<Turl> arokux: I don't know, hno, oliv3r?
<rellla> and, you don't need xserver to run xbmc
<utente> rellla, so i go with aptr-get purge xserver-xorg-etc?
<rellla> for fb blanking try echo 0 > /sys/devices/platform/disp/graphics/fb0/blank and read
<utente> ok let me try...
<rellla> you simply don't have to start it. what distribution are you using?
<utente> debina
<utente> debia
<utente> debian!
<utente> the echo didt give a solution
<utente> i suppose is enough to comment or disable some scripts... right?
<utente> hat about to change name to script /etc/init.d/x11-common?
<utente> nope, graphic start again.. what script i bave to modify to avoid X start automatically?
<rellla> dont know where xserver is started nor which watchdog is holding it up
<utente> i see /user/bin/X seems to ve the server. I try to change name to int and restart CB. hold on...
<oliv3r> Turl: arokux machine ID isn't used with DT anymore, so i think we either ignore it, or it needs something fixed
<arokux> oliv3r, you cannot ignore it, since it prevents kernel from booting
<Turl> oliv3r: I know, but there's sth fishy going on on multi_v7_defconfig in .11
<oliv3r> u-boot-spl is stored in sram-A, so at 0x00000; u-boot after that, is stored in DRAM, probably 0x400000000 I guess
<arokux> oliv3r, ok, so it's not overwritten or so
<Turl> ojn: hi, are you aware of any issues with multi_v7_defconfig preventing DT boot on 3.11-rcN?
<ojn> Turl: for what hardware?
<Turl> ojn: we're in the sunxi channel, so sunxi :)
utente has quit [Quit: Sto andando via]
<ojn> it booted fine on the two systems i have setup for automatic testing, a panda and a tegra seaboard. I don't have any sunxi hardware in the board farm yet (will hopefully add the a31 soon)
<ojn> Turl: mripard is probably a better person to ask. i'm not sure if we have enabled all the right drivers in the multi config for sunxi yet.
<Turl> ojn: are you booting multi_v7_defconfig with bootm triplet?
<Turl> ojn: it's not a matter of drivers, but we're getting things like "Error: unrecognized/unsupported machine ID (r1 = 0x00001008)
<Turl> If I disable all of the other nonsunxi platforms it works fine
Offshore has quit [Ping timeout: 240 seconds]
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
<arokux> this is the place in kernel from where the error is reported
<hno> arokux, at the top of memory.
<arokux> hno, uboot?
<hno> yes.
<hno> to make sure it's way off any locations you want to load kernels etc into.
<hno> why did you ask?
<arokux> because of thinking what can lead to: "Error: unrecognized/unsupported machine ID (r1 = 0x00001008)
<Turl> arokux: don't worry much, I'm pretty sure it's a kernel bug :)
<arokux> Turl, so let's find it :)
<Turl> arokux: ojn I'll try to triage it in a minute, and see if I find what's causing it
<ojn> Turl: yeah, i boot with bootm triplet on both panda and seaboard
<ojn> Turl: looks like your u-boot is still passing in a machine id. it should be 0xffffffff on dt boot
tzafrir has quit [Ping timeout: 264 seconds]
<Turl> ojn: how come it worked earlier then?
<Turl> is this a recent change?
<oliv3r> that would kinda suck :)
<ojn> turl: recent change to what, the kernel? the kernel has never handled sunxi atags booting.
<Turl> ojn: but I haven't changed uboot at all
<ojn> Looks like the value passed in is for:
<Turl> and it used to work
<arokux> Turl, i'm using u-boot-next
<oliv3r> having 2 u-boots would be annoying too '1 for normal ekrnel, 1 for mainline'
<ojn> Turl: what's the exact command you use to boot the kernel? and is u-boot showing that it's launching with a device tree?
<arokux> and: include/configs/sun4i.h:#define CONFIG_MACH_TYPE 4104
<arokux> 4104 = 0x1008
<oliv3r> we're not supposed to use machine types with mainline anymore
<Turl> ojn: let me get a boot log, sec
<ojn> (I've ordered a cubieboard 1 to add to the boot farm, but it hasn't arrived yet -- the cubieboard 2 isn't supported in mainline yet so i haven't added it in)
<Turl> ojn: command is bootm 0xkernel - 0xdt
<ojn> 0x102e? That's not the same as you said above
<ojn> Oh!
<Turl> ojn: well, I copied the error from arokux :) I'm using a different sunxi
<ojn> I wonder if nothing matched the compatible field, and thus didn't boot. That looks like it's for A13 hardware, and I don't think there's mainline support for it yet?
<Turl> ojn: the other interesting bit is that if I don't embed initramfs on the kernel, it works
<ojn> how big is the initramfs?
<Turl> (of course it explodes right after booting because there's no root, but that's another story)
<Turl> 4.7M cpio
<Turl> and I chose LZO in menuconfig
fredy has quit [Excess Flood]
<ojn> ok, no that large then. try using a different address for the device tree?
fredy has joined #linux-sunxi
<arokux> Turl, which uboot do u use, there is no such CONFIG_MACH_TYPE in uboot-next
<arokux> ojn, I've tried with different addresses
<Turl> ojn: I don't think it's the addresses, they're quite separated and they've been working fine for quite a while
bsdfox has quit [Ping timeout: 264 seconds]
<Turl> boot_tftp=tftp 0x50000000 dts/sun5i-a10s-olinuxino-micro.dtb; tftp 0x40000000 uImage; bootm 0x40000000 - 0x50000000
<arokux> not 0x102e, but 0x102a!!
<Turl> arokux: I'm using an A10S here
<arokux> include/configs/sun5i.h:#define CONFIG_MACH_TYPE 4138
<arokux> 4138 = 0x102a
utente has joined #linux-sunxi
<ojn> weird, u-boot should have launched with r1=0xffffffff
<oliv3r> arokux: afaik the 'hot' u-boot sunxi-current
<oliv3r> from
<arokux> ok, so everybody uses it here?
<Turl> ojn: doesn't like that machid apparently :)
<utente> rellla, ok, i disabled X, reboot and start xbmc. video is more fluid, from remote ssh i see xbmc dont use more than 15% of cpu. sometime still pink/white bug. audio is ok. now i will it run for all the movie, to see if there is some block like before.
\\Mr_C\\ has quit [Ping timeout: 240 seconds]
<arokux> Turl, so it got stuck?
<Turl> arokux: yes
<arokux> Turl, haven't you added earlyprintk?
<Turl> arokux: yes :)
<rellla> utente: which xbmc do you use? and are you using libhybris?
<rellla> *xbmc commit
<oliv3r> arokux: i use :)
<oliv3r> well that url sorta :p
<utente> rellla, im using the fixblue2
<utente> rellla, im not yusing libhybris.
<rellla> utente: do it!
<rellla> artefact bugs should be solved with some avc1 files
<arokux> Turl, weird.. it seems to me that its kernels fault
<utente> rellla, ok, but fist i prefere to leave run all movie... i use it ase test, but never seen it tille end :)
<utente> and is a goot test to see stabolity of xbmc
<arokux> Turl, you see: if mdesc is NULL it falls through here
<arokux> but it shouldn't
<arokux> Turl, something went wrong in setup_machine_fdt
<oliv3r> smells like xbmc mk802
<mripard> Turl: having problems?
<Turl> mripard: yeah, multi_v7 doesn't work when there's an initramfs :) something fishy
<Turl> mripard: on 3.11 that is
<mripard> odd
<mripard> I'm not using multi_v7, but something alike
<mripard> what's the error?
<Turl> mripard: machine id not supported
<mripard> did you change u-boot, or your u-boot script somehow?
<mripard> because it looks more like u-boot not giving the 0xffffffff machine id to u-boot when booting dt
<arokux> mripard, Turl also tried setting machid in uboot explicitly.
<arokux> mripard,
<Turl> mripard: nah, I haven't touched uboot in ages
<Turl> mripard: and it works if I don't embed an initramfs
<ojn> mripard: the other option is that the compatible doesn't match for some reason
<ojn> i.e. corrupted device tree
<mripard> ojn: doesn't it boot on the generic DT machine Arnd pushed these days?
<mripard> when it has a compatible it doesn't recognise?
<ojn> Oh, maybe it does
<ojn> but yeah, even if it doesn't recognize a compatible field it should never even have seen that r1 value
<ojn> so something's weird.
<Turl> If I actually force machid ffffffff on uboot, the kernel hangs, even with earlyprintk
<mripard> where does it hangs?
<Turl> mripard: Uncompressing Linux... done, booting the kernel.
<arokux> from what I have sofar setup_machine_fdt(unsigned int dt_phys) is called with NULL
<arokux> this causes the problem eventually
<mripard> Turl: on your A10s Olinuxino?
<Turl> mripard: yes
<mripard> arokux: which board are you using?
<arokux> ojn, if mdesc is NULL it falls through here
<arokux> but it shouldn't
<arokux> mripard, Mele A1000
<mripard> arokux: there's no DT for it so far, did you write your own?
<arokux> mdesc is NULL because setup_machine_fdt is called with NULL as parameter
<arokux> mripard, I take one from cubieboard
<arokux> :)
<arokux> (just for start)
<arokux> setup_machine_fdt gets __atags_pointer as parameter, which is NULL
<arokux> but should be a pointer to FTD, or?
<mripard> could you paste your bootlog and your u-boot environment?
<mripard> Turl: 2s, I'm committing something and try to boot mine
<Turl> mripard: ok
<Turl> mripard: do you want my env too?
<mripard> but I'm pretty sure I got it running on top of 3.11, since I wrote the clocks for it, and tested oliv3r's patches on it
<mripard> yep, please
<Turl> mripard:
<arokux> -- boot log
<Turl> mripard: test with the same branch arokux is using, it's one from your github I think
<mripard> Turl: old android habits, heh? :)
<Turl> mripard: eh? :)
<mripard> no_console_suspend ;S
<mripard> ;)
<Turl> mripard: ah :P
<arokux> -- uboot just before calling bootm
<Turl> mripard: that's because I'm writing PM and I need something to wake my board :)
<Turl> mripard: we have no other peripheral support that can wake it :(
<mripard> Turl: ah, yes
<Turl> and that's why I was hinting at the buttons thing driver today :)
<mripard> arokux: can you pass the cmdline like that as an argument to bootm?
<arokux> mripard, I've did so, didn't I?
<arokux> bootm 0x49000000 - 0x46000000 cmdline earlyprintk console=ttyS0,115200
<mripard> yeah, but I've never seen this construct before :)
<arokux> well... this is from: help bootm
<arokux> :)
<arokux> mripard, how do you want I should add cmdline?
<mripard> I usually use the bootargs env variable
<arokux> mripard, same shit. nothing changed
<Turl> mripard: are you using your wiki branch? with multi_v7_defconfig + initramfs?
<Turl> I see that you use a separate ramdisk
<mripard> it's using the sunxi-next-test-sid-next branch I have on github
<Turl> mripard: if I boot with no ramdisk embedded it works (except for the obvious "no root" death)
<Turl> mripard: can you enable initramfs embedded on the uImage and try again?
<mripard> that's the setup I'm using on the A31 and it works fine as well
<mripard> well, here at least
<Turl> mripard: is it as fresh as the sid branch?
<arokux> mripard, but you have initramfs already, right?
<mripard> Turl: fresher actually :)
<arokux> (built-in)
<mripard> arokux: for the A31 yes, for the others, no
<Turl> mripard: I'm using a ~2day max torvalds/master based branch
<Turl> mripard: how big is your cpio?
<arokux> mripard, for the log you've just shown
<mripard> that's the question I was going to ask you
<mripard> 4.1M
<Turl> mine is a 4.7M buildroot
<mripard> you know what, 2s
<mripard> I'm pushing my kernel image
<mripard> test it on your board
<mripard> we'll start from there
<Turl> mripard: with initramfs embedded? :)
rellla has quit [Remote host closed the connection]
<arokux> are you guys reading my analysis?
<mripard> nah, but we don't really care at this point to know if the kernel boots completely or not, do we ? :)
<Turl> mripard: I can boot just fine if I don't embed anything
<Turl> mripard: but then of course it's no use :P
<arokux> I've just proven runtime that: __atags_pointer is NULL
<mripard> arokux: you do realise I only have one brain?
<mripard> and __atags_pointer is whatever is filled by u-boot in r2
<arokux> mripard, Turl: guess what? if initramfs is not there, then this pointer is not NULL
<arokux> Turl, any idea where u-boot fills r2?
<Turl> arokux: nope, oliv3r, hno?
<hno> what?
<arokux> hno: __atags_pointer is whatever is filled by u-boot in r2
<arokux> any idea where u-boot fills r2?
tzafrir has joined #linux-sunxi
<hno> ah, atags pointer. It's in arch/arm/lib/bootm.c
<mripard> now, I'm confused. Turl's problem is when initramfs is embedded in the kernel, and arokux is when the initramfs *isn't* ?
<hno> both works for me.
<arokux> mripard, no, if ALSO embedded
<hno> arokux, you have initrd both embedded and separate at the same time?
<arokux> no, embedded only
<hno> and the problem is?
<arokux> hno, but i do have two files on mmc0
<arokux> the problem is it does not boot
<arokux> two files = two versions of uImage
<hno> what does it say?
<arokux> hno, I've tracked it down to the fact that __atags_pointer is NULL
<arokux> mripard, said it is whatever uboot places at r2
<hno> atags_pointer is supposed to be NULL when using DT I think.
<arokux> I checked: if no initramfs is embedded into uImage, then this pointer is not NULL
<arokux> hno, no, as far as I understand atags_pointer points to DT in memory
<mripard> hno: nah, it's 0xffffffff
<arokux> mripard, ?
<mripard> ah, nevermind. Brainfart.
<mripard> r1 is 0xffffffff
<mripard> r2 is the pointer to the DT in memory
<mripard> but it's not supposed to be NULL anyway.
<arokux> mripard, i tested it, and if initramfs is builin, it's NULL
<hno> u-boot do not know about builtin initramfs. sounds like the kenel forgets it somewhere..
<utente> hi, im revompiling libhybris to fix xbmc. one istruntions say to run ./ --prefix=/usr/local/hybris but it stop reporting
<utente> Unrecognized character \xFF; marked by <-- HERE after <-- HERE near column 1 at /usr/share/autoconf/Autom4te/ line 1.
<utente> Compilation failed in require at /usr/share/autoconf/Autom4te/ line 19.
<utente> BEGIN failed--compilation aborted at /usr/bin/autoreconf line 39.
<utente> Compilation failed in require at /usr/bin/autoreconf line 39.
<utente> BEGIN failed--compilation aborted at /usr/share/autoconf/Autom4te/ line 19.
<utente> htat file Channeldefs.ob is filled of yyyyyyyyy....
<mripard> this is with sunxi-next, I'll try with master
<Turl> mripard: apparently it's something with the config that's triggering the bug
<Turl> old multi_v7 work fine :/
tinti_ has quit [Quit: Leaving]
<mripard> here's the config
<utente> i try to reisntall mut i get error:
<mripard> Turl: what's yours?
<Turl> mripard: "make multi_v7_defconfig" + choose initramfs
<utente> dpkg: unrecoverable error: package "cmake-data" miss terminal "newline".
<Turl> (and debug_ll stuff to see the errors)
<Turl> mripard:
<mripard> ah, so you're saying it's ojn's fault ? :)
<Turl> I dunno if it is his, maybe his commit is exposing someone else's fault
<Turl> that commit is pretty big though, not so easy to know exactly what is causing it
<arokux> Turl, yes, I hate those merge-like commits
<arokux> arm: multi_v7_defconfig: Enable initrd/initramfs support
<Turl> arokux: git checkout c12d82b arch/arm/configs/multi_v7_defconfig
<mripard> that looks suspicious :)
<arokux> mripard, c12d82b: +CONFIG_BLK_DEV_INITRD=y
<mripard> arokux: I have that in my configuration as well
<arokux> mripard, Turl: update, mripard's config works for me
<Turl> mripard: it isn't the appended dtb option :(
<mripard> Turl: damn :(
<mripard> Turl: and I do have the same problem.
<arokux> mripard, with default config?
<mripard> yep.
<mripard> and current master
<mripard> but why the passed machine id is not the DT one
<arokux> mripard, :)
<mripard> and most surpringly, why is it the *right* one if we were not using DT
<arokux> mripard, can I explain findings once again?
<mripard> yep, go ahead
<arokux> setup_machine_fdt gets NULL
<arokux> thats why it returns NULL in row 197
<arokux> arch/arm/kernel/devtree.c
<arokux> arch/arm/kernel/setup.c:858
<Turl> mripard: I think some early assembler is clobbering them
<Turl> mripard: disable CONFIG_SMP and try to boot
<arokux> mdesc is now NULL so setup_machine_tags is called
<arokux> but it's not supposed to be called, I think
<mripard> Turl: CONFIG_SMP is enabled on both multi_v7 and my config file
<mripard> arokux: yep, your analysis looks right
<Turl> mripard: yeah but your config doesn'thave all the machines
<Turl> so a smp .S for, say, tegra or whatever might be clobbering stuff. It's a long shot theory though :)
<mripard> it has already two of them, and picks the right one.
<mripard> ah, yes, maybe
<mripard> it's quite late here, I'm going to bed
<arokux> mripard, __atags_pointer is NULL, but this is wrong. this is 100%
<arokux> so why it's NULL if initramfs is there...?
<mripard> let me know if you find something
<Turl> mripard: I will
<arokux> Turl, what assembly do you mean?
\\Mr_C\\ has joined #linux-sunxi
<arokux> Turl, it seems atags_pointer is set in arch/arm/kernel/head-common.S
<Turl> ojn: totally unrelated, but it caught my eye when turning off SMP
<ojn> turl: thanks, easy fix
<arokux> Turl, any ideas how to proceed?
<mturquette> Turl: ping
<mturquette> Turl: so I think that your code needs to be made into a device driver
<mturquette> Turl: and then it can have a nice DT node to fetch clocks from
<Turl> mturquette: but it'd be abusing DT then, wouldn't it?
<Turl> because "PM" is not a thing in itself
<mturquette> Turl: there has been some discussion about having a miscellaneous /drivers/syscon for handling SoC-level PM and integration needs
<mturquette> not just PM, but the rest of the chip glue that doesn't fit easily into existing driver classes
<Turl> mturquette: was this discussion on a list?
<Turl> mturquette: apart from that, I also wanted to ask you about cpufreq, what would be the best way to approach it when you need to manage multiple clocks?
<slapin_nb> AYIIIIEEEEEE!!!! it seems my u-boot patch was applied (and broke lots of things, which were then fixed) so we have good working MTD (which might be slightly updated up to taste)
* slapin_nb pushes his table with his head
<Turl> slapin_nb: upstream?
<Turl> congrats :)
<arokux> slapin_nb, is this for sunXi NAND?
<rz2k> slapin_nb: would be nice to see your linux/u-boot trees with your patches for MTD
<slapin_nb> Turl: yes, upstream
<slapin_nb> rz2k, arokux: it was important update of u-boot's MTD system to support current large NANDs, so yes, it is important for sunXi.
<slapin_nb> rz2k: as for my patches, it is all messy now. github/slapin/uboot-allwinner, but I can't prevent your disappointment - it is yours now.
naobsd has quit [Quit: Page closed]
<arokux> good night ppl
BJfreeman has quit [Read error: Connection reset by peer]