hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
popolon has quit [Quit: Quitte]
kiwasabi has joined #linux-sunxi
<kiwasabi> hey I'm having problems with the rooted Android OS images for Cubieboard2, they seem to be corrupted
techn_ has quit [Read error: Connection reset by peer]
leviathanch has quit [Ping timeout: 264 seconds]
leviathanch has joined #linux-sunxi
vinifr has joined #linux-sunxi
arokux2 has quit [Quit: Leaving]
<Turl> hi ojn
<ojn> hi
hipboi has joined #linux-sunxi
<Turl> ojn: hm, noticed too late, I already replied by mail :)
<ojn> what email?
<Turl> ojn: "[PATCH] memory: add a basic OF-based memory driver"
<ojn> Turl: Ah. there's no way to map your irc handle back to your real identity, so I didn't know that was you.
<ojn> so, that is for sunxi?
<Turl> ojn: yeah, I might need to change it one day, people often get confused
<Turl> ojn: yes, for sunxi
egbert has quit [Read error: Operation timed out]
<ojn> and which clock is it in this case that needs to stay on? the bindings says pll5 1 but that seems to be fictional?
egbert has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<Turl> ojn: the ddr output of pll5, it's not that fictional
<Turl> ojn: the first email on that chain (the RFC) has an ascii drawing of the clocks and explains the problem on detail
<ojn> Ah, but pll5 isn't in the current device tree for any of the sunxi platforms. where are the pieces for that?
<ojn> do you have all these patches in a series or on a branch somewhere, so i can get the complete picture?
<Turl> ojn: I have a ~1w old snapshot of the patches on my bitbucket
akaizen has quit [Remote host closed the connection]
<ojn> Turl: look at how tegra does this. They have a very simple init table, that takes clock names, a parent, a frequency and an enable/disable bit. For cases like these, where there will always be a need to put a reference on this one clock, there's really zero gain in trying to do it through device tree. You might as well make the clock driver aware of pll5_ddr
<ojn> and take a reference to it so it'll never be disabled
<Turl> ojn: got a path so I can take a look?
<ojn> drivers/clk/tegra/clk-tegra114.c for example
<ojn> note that you probably don't have to create a table, if this is just the first clock
<ojn> they do it for a handful of clocks, to configure their initial mux tree
<ojn> I would literally just take a reference to the clock in sunxi_init_clocks() in this case
<ojn> mturquette might disagree though, so let's have his irc tab go blue.
<Turl> ojn: so what you're basically proposing is this same patch, but stashed inside the clock driver and with the clk name hardcoded
<ojn> well, yes, instead of defining some new "simple-memory-controller" concept, which seems extremely contrived just to solve this one case that probably should just be handled in C instead of in DT.
<Turl> ojn: I took the idea from the other simple-* drivers available on the kernel, which are basically drivers for preconfigured hardware
<ojn> yeah, I think we're on a slippery slope for a bunch of then, unfortunately.
<Turl> ojn: admittedly, the user functionality gained with s-m-c is nil; others can come in handy (like simple-fb)
<ojn> yes
<ojn> exactly
<ojn> they actually configure and drive some sort of hardware
<Turl> well, except the configure part, that's the point of them isn't it?
<ojn> well, so far every single soc has had the same clock name for the ddr clock, so i don't think i'll lose sleep over that
<ojn> in particular, there is no difference at all between different boards with the same soc, which is really the most important part of device tree; to be able to add new boards without c changes
<ojn> (extra credit is being able to add new socs, but that gets hard due to pinctrl, clk drivers etc)
<Turl> hm, almost midnight already
<Turl> good night ojn
<Turl> I'll check the irc log tomorrow in case mturquette shows up
akaizen has joined #linux-sunxi
vinifr has quit [Quit: Saindo]
kiwasabi has quit [Quit: Page closed]
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
leviathanch has quit [Quit: No Ping reply in 180 seconds.]
leviathanch has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
diego71 has quit [Read error: Operation timed out]
diego71 has joined #linux-sunxi
<JohnDoe_71Rus> try to connect ttyUSB0 im minicom
<JohnDoe_71Rus> what change?
atiti has quit [Ping timeout: 264 seconds]
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
rz2k has joined #linux-sunxi
shineworld has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
shineworld has joined #linux-sunxi
Guest48558 has quit [Read error: Connection reset by peer]
soul has joined #linux-sunxi
soul is now known as Guest89329
atiti has joined #linux-sunxi
n01 has joined #linux-sunxi
rellla has joined #linux-sunxi
<mripard> vini
<mripard> rah, damn
<JohnDoe_71Rus> Veni, vidi, vici
<mripard> Turl: no, it doesn't
hipboi_ has joined #linux-sunxi
FDCX_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 260 seconds]
ssvb has quit [Quit: Leaving]
massi_ has joined #linux-sunxi
<oliv3r> morning!
<HeHoPMaJIeH> Morning
akaizen has quit [Remote host closed the connection]
<steev> the cubieboard2 should be here later! i hope, it says "delivered to unit" but it was at 5:40AM yesterday, and there is no way USPS delivered that early ;)
akaizen has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
Black_Horseman has joined #linux-sunxi
<rellla> jemk: doesn't look too bad, right? sadly i have no suitable diplay here.
<rellla> ah. jemk not here. hm.
<leviathanch> Turl: a) the sd card answers now b) sun4i_get_mod0_factors gives a division by zero when choosing 25'000'000 for clock rate
vicenteH has quit [Ping timeout: 264 seconds]
<oliv3r> rellla: :)
<leviathanch> arokux: the sd card answers now!
<leviathanch> but sun4i_get_mod0_factors produces a division by zero exception
<rah> mripard: damn?
<leviathanch> -.-
<mripard> leviathanch: cool :)
<mripard> rah: hmmm, that wasn't meant for you :(
<rah> ok
<mripard> you happen to have the nick of a complain noise :)
<rah> oh :-)
<mripard> sorry :)
<rellla> guys, what's the wheezy package to install libdri2 ?
<leviathanch> mripard: okok, I'll fix it myself
<leviathanch> >_>
<leviathanch> I'll commit a patch to Turl
shineworld has quit [Quit: Leaving]
<mripard> leviathanch: no, I was meaning "cool, that the SD card answers", not cool that you have a division by zero
<leviathanch> mripard: ok
<leviathanch> mripard: I ment because of the complain noise
<leviathanch> or did you mean someone else?
<leviathanch> ^^
<rah> np
BluesBoy has joined #linux-sunxi
<mripard> leviathanch: it was for rah :)
<leviathanch> ah
<leviathanch> ok
<leviathanch> :-)
<rellla> ok. found.
* rellla <- first read, then ask!
boycottg00gle has joined #linux-sunxi
<wens> .sb c
panda84kde has joined #linux-sunxi
<rah> oh the confusion :-)
wingrime has joined #linux-sunxi
vicenteH has joined #linux-sunxi
<arokux> leviathanch, congrats! what was the problem?
<arokux> JohnDoe_71Rus, around? I'm not sure what your minicom error says. there were cleanup patches in usb ehci/ohci, so if you think they could be the reason we can dig further.
<JohnDoe_71Rus> arokux: ok. now fine
<arokux> JohnDoe_71Rus, what was the problem?
<JohnDoe_71Rus> do not know. reconnect the usb-ttl or restart minicom
<arokux> JohnDoe_71Rus, ok..
<leviathanch> arokux: the regulator
<leviathanch> the ocr_mask wasn't set
<leviathanch> >_>
Black_Horseman has quit [Remote host closed the connection]
<arokux> leviathanch, so mmc is functionally ready now?
tzafrir has joined #linux-sunxi
<rellla> argh. anybody can lend my a hdmi cable?
<oliv3r> rellla: sure
<oliv3r> :p
<arokux> morning oliv3r
<leviathanch> arokux: I still get a memory access timeout
<arokux> leviathanch, :( solving one problem often puts you in front of the other...
<boycottg00gle> is there some known debian/wheezy/armhf bug or some related sunxi-3.4 kernel problem (memory corruption?) leading to "error: inflate: data stream error (incorrect data check)" on git clone (via squid3 http proxy)? - trying a self-hosted build as stress test here
<boycottg00gle> other hints? tips? will kick out squid3 in the next run
<leviathanch> arokux: yes
<leviathanch> arokux: it appears as if the internal DMA of the mmc host controller is too slow while accessing
<leviathanch> or I provide the wrong address
vinifr has joined #linux-sunxi
<rellla> oliv3r: at least X starts with sunxi module, mplayer plays with libvdpau-sunxi. now i need g2d enabled for sun7i and then SEE if bunny is hopping without tearing ;) next step would be to check out the xbmc thing. maybe first with the new binary, the with open source...
<oliv3r> :D
<Turl> leviathanch: just need to drop the extra division that's making noise there right?
<leviathanch> Turl: I added some "if(!foo) return;"s before the divisions
<leviathanch> and now it goes through like a charm
<leviathanch> until the memory access of the IDMA times out
<leviathanch> >_<
<Turl> leviathanch: can you try with just removing the extra division?
<Turl> leviathanch: ie, *freq = parent_rate / div;
<leviathanch> I've gotta go now
<leviathanch> later
<leviathanch> ok?
<Turl> leviathanch: nevermind then, I think it'll explode still after having a second look
popolon has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<hno> oliv3r, what findings?
<arokux> oliv3r, may corret but I think this is it hno^
<hno> Turl, having an extra i2c bus does not mean that it is not having the gps...
<oliv3r> the gps was firstly ommitted from the cubie2 sheet
ZetaNeta has joined #linux-sunxi
<mnemoc> also in Tsvetan a10 vs a20 post
<Turl> hno: they "removed" the gps mux from what I heard
<Turl> maybe it still has it, it's allwinner after all
<mnemoc> it seems everyone always used an external serial/GPS controller instead of A10's built-in
gnufan has joined #linux-sunxi
notmart has joined #linux-sunxi
<oliv3r> mnemoc: that's because the internal one is so secret! :p
<mnemoc> oliv3r: so secret that not even AW has access to the sources of the driver
<mnemoc> THEY receive a .ko
<oliv3r> hno: that commit arokux linked i tried to setup nice defines etc for the HPCR, but i ran into some inconsistencies i dundn't undertand, so i 'took a break' from it ;)
<oliv3r> mnemoc: yeah
<oliv3r> a little birdy also told me, the next AW SoC will feature dual channel memory controller, 4x4 (big.LITTLE) stuff most likly a powerVR
<oliv3r> and they haven't and probably won't buy the linux blob from powervr
<mnemoc> another pvr? :'(
<oliv3r> well they bought powerVR IP blocks
<mnemoc> lkcl_ said that allwinner's god had promised no more pvrs....
<oliv3r> so they will use it
<oliv3r> well that probably was allready in design by that time
<libv> what would be wrong with mali-450?
<oliv3r> back then, they had mali mp1
<oliv3r> so they bought powerVR due to performance reasons
<oliv3r> they got burned and regret it
<oliv3r> but they also spent a lot of money on it
<libv> or... being able to clock the mali up properly
<wingrime> libv: there some mobile gpu that can opengl , with opengles
<oliv3r> so A31 v2 will user powervr still, as it's the same chip, just replace quad core for 4x4
<oliv3r> libv: also a proper memory controller would help ;)
rellla has quit [Quit: Nettalk6 -]
<oliv3r> mnemoc: anyway, a drop in replacement for a31 is the next okt. soc
<mnemoc> :\
<mnemoc> my dreams for a a20 successor vanished
<arokux> next okt. you mean in a month?
<arokux> oliv3r, ^
<oliv3r> well whenever it was announced
<oliv3r> or would be
<oliv3r> i read on ML that it would be 4th quarter?
<wingrime> mnemoc: interesting that our mem contoller can also have ability to disable host
<oliv3r> or oct?
<oliv3r> mnemoc: there will be an a20 success
<oliv3r> successor, they aren't going all powerVR i don't think
<wingrime> mnemoc: so, trustzone , and propiraty stuff can also be disabled form dram access
<mnemoc> wingrime: :o
<wingrime> mnemoc: together with prioriy change ability
<oliv3r> wingrime: did you see my github commit? :)
<wingrime> oliv3r: url
<oliv3r> wingrime: might make things a little easier
<oliv3r> what i fail to understand, sun7i seems to be more like sun5i then sun4i, and that's fine, but why does sun4i have more HPCR's then sun7i
<oliv3r> and why did they change ...
<arokux> mnemoc, does this one will work for A10? modules/gps/gps.ko_
<oliv3r> arokux: in theory, it's for a10
<mnemoc> but 2.6.36....
<mnemoc> so REing is needed
<wingrime> mnemoc: that ko was buildable?
<mnemoc> wingrime: no, AW receives it as blob
<mnemoc> but symbols are included
<oliv3r> TJ- from replicant did look at it, he's currently REing the samsung ones
<arokux> mnemoc, but if there is no external gps module connected then testing is difficult, right?
<mnemoc> and it's marked as GPL internally....
<wingrime> mnemoc: but thats difficult to link
<wingrime> mnemoc: something in ko
<mnemoc> arokux: it needs a RF companion chip from what I know
<mnemoc> like the PHY for the ethernet
<wingrime> mnemoc: do you know GPS IP vendor , or can ask benn?
Guest89329 has quit [Read error: No route to host]
<mnemoc> wingrime: it's on the commit text
<wingrime> mnemoc: link please
<arokux> wingrime, ^
<wingrime> arokux: thanks
<arokux> mnemoc, but we do not even know how this RF can be connected to a soc and how to mux pins, right?
<wingrime> mnemoc: I have ida why it secret
<wingrime> mnemoc: do you heared that china maps are shifted by secret formula ?
<arokux> wingrime, :D
<mnemoc> doh
<mnemoc> arokux: yes, the pins and muxes are known
<libv> ssvb left us :(
<libv> he doesn't like us anymore :(
<mnemoc> ???????
Soru_ has joined #linux-sunxi
<libv> (or maybe he just fell off irc for a bit :))
<mnemoc> but I showered today!
<libv> :)
<arokux> mnemoc, so is there a board then with an RF that uses the on-chip GPS?
<mnemoc> no, but there is gps info in the template .fex files
<oliv3r> you ... shower? eww thou heathen!
<wingrime> mnemoc: sun3i also use fex
<wingrime> mnemoc: you can notice we have sections
<wingrime> mnemoc: that we have no in linux
<oliv3r> wingrime: probably even sun2i did :)
<oliv3r> wingrime: there's even sunxi extensions, fex sections we added :)
<wingrime> mnemoc: can you ask benn about get their old rtos
<mnemoc> wingrime: the only reason to "ignore" sun3i is that we don't have hardware to test. they still sell a lot of them (with RTOS) and even release new SoCs
<oliv3r> wingrime: sample GPS hardware would be more interesting, so we know what to conjnect, how and how to test it
<oliv3r> Tsvetan: ^ allwinner A10, A20 GPS module for olimexino
<wingrime> oliv3r: modern gps IC have uart
<wingrime> oliv3r: so its not a problem
<oliv3r> mnemoc: alot being relative, Tsvetan said that a SoC vendor sells 90% in the first few months
<oliv3r> wingrime: yeah, but again, TJ- from replicant was looking at the GPS module a while back; just pinged him a message
<wingrime> mnemoc: can you ask benn get old melis images/any old binnaries/sources
<oliv3r> wingrime: but what do you want with that old codebase?
<oliv3r> it's not GPLed probably
<wingrime> oliv3r: RE
<oliv3r> wingrime: maybe it could help, but I doubt it, we only need to RE the GPS
<oliv3r> which, if TJ- succeeds, should be relativly easy
<oliv3r> also, how can you transform a GPS?
<oliv3r> you take a photo of the area, you have GPS signals
<oliv3r> it's not something they can influence really
<oliv3r> There are rumor that all printed maps that is available in domestic market are transformed. The goal is to make the map useless for military situation to protect the safety of the country.
<oliv3r> PRINTED maps you can do weird shit with
<oliv3r> arokux: it's A10 though
<arokux> oliv3r, in menu there are A20 and A31
<wingrime> mnemoc: thats lie , that aw have no sourecs
<wingrime> oliv3r: ^
<oliv3r> arokux: ah ok cool
<wingrime> 'lnxdrv_F30.c'
<oliv3r> wingrime: maybe Allwinner SENT hat file to highvision, and they compiled it?
<oliv3r> wingrime: but the transposition, would only be an issue for the maps + gps signals, pure gps just outputs true coordinates
<oliv3r> so i can imagine, that the GPS modifies the coordinates to match the map
FDCX_ has quit [Ping timeout: 260 seconds]
<oliv3r> maps based on GPS or sattelite imagerary, like openstreetmap should 'just work' with normal GPS signals
<Tsvetan> oliv3r we even do not know what inside A1X is called GPS
<wingrime> BL HivisionGpsRfCheck
hipboi_ has quit [Quit: Leaving]
<wingrime> oliv3r: it can be vendor locked
<Tsvetan> most GPS chips I have opened/viewed are LNA + receiver + ARM7
<Tsvetan> what part of these is inside A1x nobody knows
<Tsvetan> do it need just antenna? or antenna + LNA or antenna + LNA + receiver
<Tsvetan> also the firmware of these is encrypted and mortals have no access to it, as all restrictions for the commercial GPS are in the firmware
<oliv3r> Tsvetan: yeah, but OUTside bits, so it could be RE-ed
<wingrime> Tsvetan:
<mnemoc> wingrime: hipboi told us on 2.6.36 times that they send the SDK to that company and receive the .ko back from them
<mnemoc> wingrime: it might have changed with the years
<wingrime> mnemoc: I can get some info from blob
<oliv3r> Tsvetan: i'm supprised allwinner won't tell you what you need to connect on the outside :)
<wingrime> mnemoc: its not big
<mnemoc> wingrime: but at that time, that is what hipboi said
<Tsvetan> oliv3r never asked them as I think this is not interesting
<mnemoc> wingrime: and has symbols
<oliv3r> IF someone would want to RE it, they should to to TJ- from replicant, he knows everything about GPS it seems
<oliv3r> Tsvetan: i agree it's one of the lowest priortiy things
<oliv3r> the symbols will help hugely
<oliv3r> but i agree with Tsvetan it's one of the last things to worry about
<arokux> oliv3r, what are the other things we do not know about? :)
<Tsvetan> BTW with these military stuff things, two days ago here was TV film for the 11/09 the atentates against the twin towers
<Tsvetan> they said that the secure encrypted channel stop working when this happend
<Tsvetan> and they had to use normal GSM phones for communication
<oliv3r> arokux: that's pretty much it, i think the only blobs are GPU, VPU and GPS
<oliv3r> we mere humans get screwed over by all the militairy powers from everywhere
<wingrime> Tsvetan: fronted GPS ic have name HV5820
<Tsvetan> friend of mine who was in the Iraq also told me that their military GPS system was always out of order and they had to buy commercial GPS-es to may know where they are
<mnemoc> doh
<Tsvetan> so military use the civil communication systems as they seems more reliable :)))
<wingrime> mnemoc: can't find datashit for HV5820 :(
<arokux> are their some known usage of the on-chip GPS IP?
<arokux> there*
<wingrime> oliv3r: take a look
<Tsvetan> wingrime there are GPS/GLONAS GPS complete chips which cost $1.50
<popolon> wingrime, I don't use GPS :)
<wingrime> yes, and with uart output
<popolon> I prefer to play with my intuition, that's a funny game
<Tsvetan> why you will use some weird front end IC to use some undocumented A1x chip feature
<wingrime> Tsvetan: see shots
<arokux> wingrime, there is something like this:
<popolon> there are chips with GPS+glonass+beidu too (and + gallileo+.....)
<arokux> but won't load
<popolon> s/beidu/COMPASS/
<oliv3r> Tsvetan: i thought the signal was the same, but you have to decrypt the 'extra digits' to get more accurate location
<wingrime> Tsvetan: thats worked , at least on sun3i
<oliv3r> arokux: tablets/carputers love GPS
<wingrime> oliv3r: see images
<wingrime> oliv3r: thats f20
<oliv3r> yeah looking right now
<oliv3r> F20 is sun3i isn't it
<mnemoc> yes
<oliv3r> the pinouts might mat the A10
<oliv3r> a10 has 'MAGN, SIGN and 1 more i forgot'
<oliv3r> though RX_EN and OSC_EN i don't think
<oliv3r> CLK might be the 3rd GPS pin A10 has
<vinifr> mripard, ping
<arokux> what part of the "GPS job" might the on-chip GPS IP do?
<wingrime> arokux: math
<arokux> maybe we need to check if there is gps.ko_ in some of the AXXx based tablets, sun3i isn't interesting from what I understand
<arokux> Turl, where did you get if from? ^
<oliv3r> arokux: sun3i is old, slow and nobody has one, they still sell them, they are used in e-readers etc, but nobody here has hardware or interest :)
<arokux> oliv3r, this was my understanding too.
<arokux> oliv3r, but see what rz2k has shown
<oliv3r> that's sun4i?
<arokux> android4.0/device/softwinner/apollo-mele/modules/modules/gps.ko
<arokux> this is from sdk-2.0
<oliv3r> maybe it's the same GPS .ko shipped?
<arokux> oliv3r, no different size
<oliv3r> if not, that's good to know that there's an updated .ko hpefully it has debug modules too
<oliv3r> if its stripped; it's less usefull
<arokux> android4.0/device/softwinner/apollo-upp/modules/modules/gps.ko
<arokux> android4.0/device/softwinner/apollo-osepp/modules/modules/gps.ko
<arokux> what are those appollo-* things?
<arokux> Phone Model: Rikomagic MK802
<arokux> Device: apollo-mele
<oliv3r> arokux: apollo might be the nickname just like nuclear was?
<mnemoc> those are usually the nicknames of the EVBs
<mnemoc> not the SoCs
wingrime has quit [Ping timeout: 264 seconds]
<arokux> mnemoc, so there is nothing interesting inside of those gps.ko?
<mnemoc> i don't know. it's probably a 3.0 compatible equivalent of the other .ko I showed before
<mnemoc> that one was for 2.6.36
deasy has joined #linux-sunxi
dlan has quit [Quit: Lost terminal]
<vinifr> Does sun4i-ts.c really works? Has someone already tested a tochscreen?
ssvb has joined #linux-sunxi
<mnemoc> yes, the guy who made the cubiebase board
dlan has joined #linux-sunxi
<rz2k> vinifr: sun4i-ts works on olimex lcd modules
<rz2k> s/on/with/
<vinifr> It is so strange :)
<vinifr> rz2k, works in Android or Linux?
<rz2k> linux
<rz2k> you need patched tslib
<rz2k> and patched xf86-input-tslib
<libv> mnemoc: ssvb still loves us!
<boycottg00gle> uh now nbd died
<libv> ssvb: the destruction crap remains in the binary, will have this fixed in a few
<mnemoc> libv: \o/
Soru_ has quit [Read error: Connection reset by peer]
soul has joined #linux-sunxi
soul is now known as Guest42969
<atsampso1> mnemoc: it might be useful to have a dictionary of Allwinner internal names somewhere...
<mnemoc> atsampso1: wiki? :)
<atsampso1> yeah
<mnemoc> go ahead!
<atsampso1> well, I would, except that I don't actually know any of them ;)
<mnemoc> registration is free ;-)
<mnemoc> iirc nuclear is a13
<mnemoc> crane is a10's EVB
<atsampso1> so does that make crane... yes
<libv> ssvb: the neutered EGL binary is in your inbox, can you run es2_info against it?
<atsampso1> so I guess "crane-anpei" is Ampe's modified version of the crane config...
<rz2k> [15:51:50] <atsampso1> mnemoc: it might be useful to have a dictionary of Allwinner internal names somewhere... -
<rz2k> please add what you know there
\\Mr_C\\ has quit []
iamfrankenstein has quit [Ping timeout: 264 seconds]
<arokux> lkcl_ won't join our ml because it is on google groups :$
<arokux> what's the point? I do not see google groups interface and use my e-mail client or gmane
<arokux> what is the matter where it is?
vinifr has quit [Quit: Saindo]
<arokux> it --> our ML
<oliv3r> arokux: not really sure, but we will switch to our own mailing server at some point anyway
\\Mr_C\\ has joined #linux-sunxi
<arokux> oliv3r, yes, I remember mnemoc promised to set it up.
<libv> arokux: it's not like that is a burning matter, now is it
<arokux> not at all.
<arokux> libv, that is why I was surprised he won't join a list because it's on google groups
<arokux> anyway, his style and behavior is a bit ... not of a *usual* person :)
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier.]
<mnemoc> arokux: initially the ML issue was more complecated because we wanted to host CB's ML too .... and mailman is not good at dealing with multiple domains
<atsampso1> huh, I didn't realise Allwinner had their own wiki!
<mnemoc> but with CB moved to china it should be pretty trivial to install postfix/mailman on the sunxi server
<mnemoc> Turl: ----^ ? :)
Turl|web has joined #linux-sunxi
<arokux> I regret I've missed a split of linux-sunxi from arm-netbook
<Turl|web> mnemoc: I hate mail servers :)
<mnemoc> Turl|web: :(
<arokux> I do not bother where our ML list is, as I do not see its "location" anyway.
<mnemoc> sometimes G damages the patches
<mnemoc> for others, it's a religious problem
<boycottg00gle> my mails are delayed for ever by g00gle
<arokux> boycottg00gle, ah, now I know why you have the nickname you have
Guest42969 has quit [Remote host closed the connection]
<boycottg00gle> arokux: nah it is more about prism
<Turl|web> boycottg00gle: you should rename yourself to boycottnsa then
<arokux> Turl|web, wanted to say the same!!
<arokux> wrong link
<arokux> you'd need to put it in google translate yourself, unless you are not religious
<boycottg00gle> nsa - google < epsilon
<ynezz> hm, and who cares?
<boycottg00gle> too few i guess
<ynezz> once it's digital, it's defacto public, if you think otherwise you just don't get it
Turl|web has quit [Quit: Page closed]
<ynezz> and it's quite offtopic here
<mnemoc> if you want privacy, move to live to a cave in an isolated mountain
<boycottg00gle> i stop here - off-topic ...
<mnemoc> ack
<arokux> Tsvetan, do you think sdk-2.0 includes this patch?
<oliv3r> mnemoc: on what server do you want me to install postfix :)
<mnemoc> oliv3r:
<mnemoc> oliv3r: do you have sudo there? it's a ubuntu 12.04
<mnemoc> only for aliases and ML, no private mailboxes
<mnemoc> wanna take the job?
<mnemoc> forwards for community members, smtp auth, and dev@ ML
<oliv3r> mnemoc: i think so yeah
<mnemoc> \o/
<oliv3r> mnemoc: i'll do it when i have time :p which may be today, but might be next week
<oliv3r> but yeha i'll look into it
<oliv3r> i'll do the most simple basic setup without many complex things
<mnemoc> amazing. thank you!
<oliv3r> well you never asked
<oliv3r> you only ask your friend turl!
<atsampso1> has anybody actually used the sun4i_csi camera drivers lately?
<mnemoc> oliv3r: I bother him becase the last time he offered himself to help some sysadmin-ish issues in the server
<atsampso1> all the device modules #define I2C_ADDR, but none of them use it, so unless you explicitly tell them with an insmod parameter they won't know what I2C address to use...
<mnemoc> oliv3r: now I know I can bother two community members with sysadmin-ish tasks :p
<atsampso1> I assume this wasn't true at some point in the past, since my tablet came with gc0308.ko that doesn't need a parameter
<arokux> mnemoc, who's donating a server? :)
<atsampso1> interestingly, it also came with gc0308b.ko for the second camera, which after some peering at objdump output is just gc0308 with a different I2C address hardcoded into it ;)
<mnemoc> atsampso1: the driver might be modified to take arguments from script.bin
<mnemoc> arokux: the current one, a friend of mine from chile
<mnemoc> the previous 2, me
<mnemoc> but this one now is permanent
<arokux> has address and traces lead to
<arokux> which is in germany
<mnemoc> best hosting on earth :)
<arokux> mnemoc, why? :)
<mnemoc> price/quality/service
Soru has joined #linux-sunxi
<mnemoc> best hardware for the money, and very good support, bw, and connectivity
<arokux> mnemoc, bw?
<mnemoc> bandwidth/speed
<arokux> mnemoc, so our server is hosted there? what happens if their HDD crashes?
<mnemoc> it has 4, in raid10
<specing> they replace the HDD?
<mnemoc> yes
<specing> that was a rethorical question
<mnemoc> you give the serial of the defective disk and they replace it within the hour
<arokux> mnemoc, just wanted to know how backup-ing works
<specing> they have someone lurking in there at night?
<specing> anyway, I'd like to get hold of one of those kimsufi 3 Eur/month dedicated Atom boxes
<mnemoc> arokux: none setup yet. but we can request a 100GB ftp account
<specing> rsync to home?
<arokux> mnemoc, but if raid10 then none is needed?
<mnemoc> arokux: well... the DC can be bombed and the 4 drives die at the same time
<atsampso1> mnemoc: that's not a bad idea, but it's not been done yet ;)
<mnemoc> :)
<atsampso1> ... that said...
* atsampso1 looks at the right bit of code ;)
vicenteH has quit [Ping timeout: 246 seconds]
<mnemoc> arokux: one of each group can die. so up to two can die..... but if both of the same raid1 group die, you are screwed
<mnemoc> fortunatelly chances of that happening "naturally" are slim
<arokux> mnemoc, yes, that's why I said 2
<mnemoc> but setting up proper external backup is in the TODO
<mnemoc> hetzner gives 100GB over ftp for that
* atsampso1 peers suspiciously at the .fex file, which has the different I2C address for the second camera (0x46) from the one hardcoded into the gc0308b driver (0x21)
* atsampso1 wonders if that's why they wound up hacking the driver...
<arokux> mnemoc, so is one of the "root servers"?
<mnemoc> and lxc inside a root server
<mnemoc> an lxc inside a root server
<mnemoc> with dedicated IP
<arokux> mnemoc, why for lxc?
<boycottg00gle> mnemoc: we had recently many seagates 3tb failing @ hetzner (in one box 2 of 2)
<rz2k> lxc==lxr?
<mnemoc> lxc is mainline implementation of linux-vserver
<mnemoc> still incomplete for hosting, but good enough for own server isolation
<rz2k> oh
<rz2k> ok
<rz2k> what about lxr anyways?
<arokux> rz2k, The Linux Cross Reference?
* rz2k votes for linux-sunxi cross-reference with 15453 hands
<mnemoc> rz2k: on the TODO. the mirrors are already there, so lxr can be configured upon those
<mnemoc> also cgit and git:// of the mirrored repos are pending
<mnemoc> too much to do and too little time
<arokux> rz2k, lxr for chinese code?
<rz2k> yes
<oliv3r> mnemoc: it's no bother :p i'll try to do it as time permits :)
<arokux> mnemoc, why have you decided to install lxc?
<rz2k> i cant read ___________type WeirdName( ton of parametes) without my eyes bleeding out
<rz2k> lxr helps a lot
<oliv3r> mnemoc: but as you, too little time, too much to do
<arokux> mnemoc, i'm not asking about why lxc and not something else, but why lxc at all.
<oliv3r> arokux: what's wrong with lxc?
<mnemoc> arokux: because it's mainline and enabled by default in ubuntu's kernel
<oliv3r> arokux: to departmentize a server can be beneficial
<oliv3r> arokux: and with linux on linux, why bother with a KVM unless you absolutly have to
<arokux> as said, i'm just interested in the question why visualization at all.
<mnemoc> vserver/lxc has no emulation, so no wasted resources if you don't need separated kernel
<oliv3r> arokux: scenario 1) i have a serve,r it runs hardened gentoo due to security concerns, it being directly connected to the internet etc
<oliv3r> arokux: but i also want to host a remote server for myself, to test things on etc, but don't want to get a 2nd server
<oliv3r> arokux: so i run something more advanced with bleeding edge package and testing random stuff
* boycottg00gle uses a small script to clone a temporary vm to test stuff
<oliv3r> arokux: or, what if i have 2 domains, unrelated to eachother, so i don't want them to have any possible way to reach eachothers files etc
<oliv3r> boycottg00gle: and that too!
<oliv3r> arokux: while I"m not a huge fan of virtualization, but I deff. see its benefits and also do use it where appropiate :)
<oliv3r> baremetal for everything else :)
<arokux> oliv3r, I see, thanks
<boycottg00gle> moving a vm is handy sometimes, too
<mnemoc> the sunxi lxc was moved with a simple `tar | gzip | ssh tar`
<n01> I'm using ubuntu inside an arch for that fuck*ng yocto
<n01> lxc saves asses
<arokux> n01, why do you use it if it is f**ing? :)
<mnemoc> employer mandate?
<mnemoc> who uses yocto for own pleasure?
<rz2k> n01: yocto doesnt play on arch?
<arokux> well, employer must be unhappy to, if it is so f**ing
<mnemoc> yocto is one of those projects that imposes a lot of restrictions/assumtions on how it has to be used
<mnemoc> but you can get used to it
<n01> mnemoc: yes, employer -.-
<n01> rz2k: nope, a lot of problems with texinfo and such
<n01> arokux: they already have something on yocto and they want to continue with that. and yes, I complained a lot
<n01> I have to push allwinner also here :)
<mnemoc> we should have meta-sunxi in our github....
<mnemoc> if someone is willing to maintain it
<arokux> mnemoc, what should be in?
<mnemoc> sunxi-specific packages
<arokux> mnemoc, for which distro?
<mnemoc> kernel, bootloader, gpu, vpu, sunxifb
<mnemoc> meta-* is openembedded (yocto)
<mnemoc> that's how you glue platforms in oe
<n01> but it's only for A10
<mnemoc> n01: i mean an "official" one and for all the socs
<mnemoc> and using 3.4
<n01> you can ask _enrico then. He was here a couple of days ago
gnufan has quit [Read error: Operation timed out]
ykchavan has joined #linux-sunxi
boycottg00gle has quit [Remote host closed the connection]
techn_ has joined #linux-sunxi
FDCX_ has joined #linux-sunxi
jemk has joined #linux-sunxi
wingrime has joined #linux-sunxi
<atsampso1> "The following text is what triggered our spam filter: We have doubts if you are a spambot or not, please <a href="/IRC">join us on IRC</a> to sort this issue out."
<atsampso1> I'm not a spambot ;-) (atsampson on the wiki)
arokux2 has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
atiti has quit [Read error: Operation timed out]
atiti has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
atsampso1 is now known as atsampson
<wingrime> atsampson: wait Trul
atiti has quit [Ping timeout: 246 seconds]
naobsd has joined #linux-sunxi
<wingrime> mnemoc: ping
<Turl> atsampso1: try again
<Turl> wingrime: type my irc name correctly for ping :P
<atsampson> Turl: yep, that worked; cheers!
n01 has quit [Remote host closed the connection]
BluesBoy- has joined #linux-sunxi
BluesBoy has quit [Ping timeout: 264 seconds]
BluesBoy- is now known as BluesBoy
panda84kde has quit [Remote host closed the connection]
<mnemoc> wingrime:
<wingrime> mnemoc: someone tryed add new cedar in ml
<wingrime> mnemoc: but its ugly
<mnemoc> comment, suggest, and request a v2
<mnemoc> and a v3,.... v4.... v5.... until it's not ugly :)
<wingrime> mnemoc: I want make it myself but after I figure whats wrong with cedar on a13
<wingrime> mnemoc: some strange regression with timer? when moved form 3.0 to 3,4
BluesBoy has quit [Quit: HydraIRC -> <- Chicks dig it]
massi_ has quit [Quit: Sto andando via]
<wingrime> mnemoc: its ugly for many resons
<mnemoc> uhm
<wingrime> mnemoc: 1) he not moved to sunxi . instead using sun4i
<wingrime> 2) only sun4i and sun7i merge
<wingrime> 3) he added sunxi_is where disable L1/L2 for cedar mem
<wingrime> thats must be removed
<wingrime> 4) old code from sun3i
<wingrime> 5) strange revisions checks ....
<wingrime> he not tryed to understand it
<wingrime> mnemoc: 8 comments about that patch
<techn_> understanding is different thing
<techn_> unless there is other stuff than merge
<wingrime> techn_: have you looks that patch?
<wingrime> *looked
<techn_> just quick look.. not with mind :)
<wingrime> vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
<wingrime> it will make regression on a10
<techn_> but sure its ugly :)
<wingrime> techn_: can you look my patches for disp
<techn_> wingrime: I have no strong opinnion of removing dead code.. It's good for documentation .. usually no one checks history if there is tips how things work
<techn_> but ofcrouse it's not always good for readability
<techn_> but anyway I have tried to preserve those until something is cleaned or "done"
<techn_> hopefully I send patch for pwm backlight during this weekend.. It's already working but it causes API break
<techn_> So I'm wondering if I should make some code to keep API.. so that current android images wont break..
<techn_> Turl: ^
<Turl> techn_: backwards compat is always nice to have
<Turl> it shouldn't be too difficult, it's just an ioctl on disp that you need to reimplement
<techn_> Turl: yes.. disp needs handle to backlight instance
<techn_> but I think we can kill disp PWM api :p
<wingrime> mnemoc: can you push?
<wingrime> techn_: I have 3 patch for disp
<wingrime> techn_: 1) remove dead blocks (as you seen)
<wingrime> techn_: 2) expose unused blobcks
<wingrime> techn_: and 3) make disp soc select fully dunamic
rz2k has quit [Read error: Connection reset by peer]
jemk has quit [Ping timeout: 260 seconds]
Soru has quit [Read error: Connection reset by peer]
atiti has joined #linux-sunxi
soul has joined #linux-sunxi
soul is now known as Guest61171
jemk has joined #linux-sunxi
sanka has joined #linux-sunxi
Soru has joined #linux-sunxi
Guest61171 has quit [Ping timeout: 264 seconds]
paulk-aldrin has joined #linux-sunxi
ykchavan has quit [Quit: Leaving]
<wingrime> error: enc_dec_test uses VFP register arguments, ./linux_lib/libcedarv.a(vc1debug.o) does not
<wingrime> anyone have idea how to deal?
CoSM0 has quit [Read error: Connection reset by peer]
CoSM0 has joined #linux-sunxi
ZetaNeta has quit [Read error: No route to host]
dwilkins has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<mnemoc> wingrime: armel vs armhf?
ZetaNeta has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 264 seconds]
tinti has quit [Read error: Connection reset by peer]
Black_Horseman has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<wingrime> mnemoc: thats problem
<wingrime> mnemoc: I want build encoder sources from aw
<wingrime> mnemoc: my system armhf
<wingrime> mnemoc: but f** blob no
<mnemoc> get an armel toolchain
<wingrime> mnemoc: with armel toolchain
<wingrime> mnemoc: I have ../../arm-linux-gnueabi/bin/ld: cannot find -lstdc++
iamfrankenstein has quit [Ping timeout: 264 seconds]
<mnemoc> get armel libstdc++ :p
<mnemoc> or.... make an armel chroot
<mnemoc> debootstrap or so
<wingrime> mnemoc: thats embebian problem
<wingrime> mnemoc: no link
<mnemoc> debootstrap a native armel in a subdir of your armhf system, and play with aw's blob in there
<wingrime> mnemoc: are you tryed run enc_dec_test?
<mnemoc> nope
<mnemoc> i'm telling you how to get an armel system within your armhf system
<mnemoc> then try that stuff within chroot
<wingrime> mnemoc: yeax, I will bootstrap it now
n01 has joined #linux-sunxi
<wingrime> mnemoc: that guy make v2
<wingrime> mnemoc: but drop comments traslations
<mnemoc> :'(
arete74 has quit [Ping timeout: 246 seconds]
Olaff has quit [Ping timeout: 246 seconds]
Olaff has joined #linux-sunxi
<mnemoc> v3 :)
<wingrime> mnemoc: not only
<wingrime> mnemoc: 1) frequency change drisabled for sun5i (I think that marketing crap)
paulk-aldrin has quit [Quit: Ex-Chat]
<wingrime> (limit deocding frequency for make it decode small video only
<mnemoc> comment on the ML :)
arete74_ has joined #linux-sunxi
<mnemoc> still need to take care of much older pending commits
<wingrime> mnemoc: lest
<wingrime> *lets
<mnemoc> wingrime: i would prefer m, but please send the commit to the ML for others to comment
FDCX_ has quit [Read error: Operation timed out]
<wingrime> mnemoc: ... i tested only "y" and emac can be used for nfs root , so I prefer y
<mnemoc> i understand the motivation for y, but most devices don't have ethernet
ZetaNeta has quit [Quit: ZetaNeta]
<arokux2> wingrime, tablets do not use emac, so M
<arokux2> wingrime, also other options required for NFS root are not set by default
<wingrime> arokux2: fex will disable it
<wingrime> arokux2: on tablet
<mnemoc> it's good to build everything sunxi, but not a built-in
<mnemoc> as
<mnemoc> if you want built-in you can always tweak the .config
<mnemoc> emac=y is only useful if nfs is also y
<mnemoc> which would lead to a never ending bloating
<arokux2> mnemoc, can we create something like sun4i_allyesdefconfig?
<arokux2> mnemoc, because I kind of can see the point of wingrime
<mnemoc> first we need consistent per-soc defconfigs for linux and android
<Turl> oliv3r: I got a suitable sd card :P
<mnemoc> but we can't enable everything
jemk has quit [Ping timeout: 246 seconds]
<mnemoc> m is fine, but everything y is like returning to the 90s
<wingrime> mnemoc: also I can 3 patch for disp from two weeks ago in ML
<wingrime> *have
<mnemoc> wingrime: please paste me the subjects in /q
<wingrime> mnemoc: [PATCH] sunxi:disp: remove bloat-code
<mnemoc> i need to leave this apartment tomorrow morning so pretty busy cleaning and packing
<mnemoc> wingrime: in /q ..... the buffer here will eat it
<mnemoc> but only the subjects. no talking :)
<wingrime> mnemoc: ?
<wingrime> mnemoc: /q in irssi? have never use it
<mnemoc> /q, /query "private message"
<wingrime> ah
<wingrime> techn_: ping
<wingrime> techn_: image_enchance_tab separation was done becose I firstly thinked that it realy must be different on sun4i and sun5i, so firstly I done separation and moved sin/cos precalulation table to another table, (shared stuff) but after I tryed on real hw
<wingrime> techn_: it was clear that it works normaly
<wingrime> techn_: also patch tested on a10 a13 and a20
<wingrime> techn_: so, thats why it in commit
<techn_> yeah.. but since it's clearly different commit, it should be separated
<techn_> wingrime: display details are hard to test
<techn_> gamma courve etc
_BJFreeman has joined #linux-sunxi
<wingrime> techn_: lcd, hdmi works nicely
<techn_> yes.. but are colours right?
<wingrime> techn_: yes
<techn_> you cant spot them on naked eye.. atleast if itsnt trained
<wingrime> techn_: ...
<techn_> .. eventually someone will come and whine about wrong colours
<techn_> like already someone did with too bright blacks
_BJFreeman is now known as BJfreeman
<wingrime> techn_: LCD not ideal
<wingrime> techn_: thats interesting, becose any phone screen and camera have own color correction magic
<wingrime> techn_: thats why iPhone and GSIII make differet colors
<wingrime> techn_: I used sun5's table
<wingrime> techn_: I think it more fresh from aw
<wingrime> techn_: but I think we have no reason use differnet tables
<wingrime> techn_: if someone comes and say that colors are wrong, lets him remake table
<wingrime> techn_: and interesing thing, if I leaved different table, sin/cos table separation defenetly commit part
<wingrime> techn_: *two different tables
<wingrime> mnemoc: are you moving and moving and moving .....
<mnemoc> wingrime: life sucks
<mnemoc> also updating CV
<wingrime> CV?
<wingrime> OpenCV ?
<mnemoc> resume, curriculum vitae
<wingrime> never heared ...
ZetaNeta has joined #linux-sunxi
<wingrime> I simply getting magister level....
leviathanch_ has joined #linux-sunxi
leviathanch has quit [Ping timeout: 240 seconds]
notmart has quit [Quit: notmart terminated!]
BJfreeman has quit [Read error: Connection reset by peer]
<wingrime> mnemoc: that table have no any reference to it in code
<wingrime> mnemoc: it simply exsist but not used
n01 has quit [Remote host closed the connection]
<wingrime> mnemoc: we have also simular table for sun4i and for sun5i
<wingrime> mnemoc: it looks was for video speical, but not implemted
notmart has joined #linux-sunxi
<wingrime> mnemoc: ?
<wingrime> mnemoc: about #if 0, we have git, code must be clean , thats why we remove sun3i isn't?
notmart has quit [Quit: notmart terminated!]
arokux2 has quit [Remote host closed the connection]
<wingrime> mnemoc: I can't split easy, that file constain tons fixes
<ssvb> libv: got wisdom teeth removed today, and it looks like the whole weekend is going to be not very productive in terms of software hacking
<mnemoc> i never agreed with the removal of sun3i. i also disagree about the utopian idea that people will look at the history to find code that was removed because we didn't know it's purpose. people looks at the history to see how current code reached it current form or to find a regression.... not for finding stuff that was there but no one cared to fix or understand
<libv> ssvb: ouch, my condolences
<mnemoc> but this is not a dictatorship, so if other devs familiar with sunxi-disp agree with you, the patches get in
<libv> ssvb: i hope that the pain will be managable
<libv> wingrime: disp will die, in a very distant future
<wingrime> libv: how do you think remorval unknown "#if 0" is good?
<ssvb> libv: as I can see from the ml, hramrach has some ideas and/or new information about fixing these egl crashes
<wingrime> ssvb: ^
<libv> wingrime: #if 0 or #ifdef UNUSED unused code
<libv> wingrime: do not throw code away
<libv> ssvb: yeah
<libv> i am currently filing for the graphics devroom at fosdem
<wingrime> libv: in case #UNUSED, I mostly make it enabled
<libv> i just got the multiple limare test working properly
<libv> i will look into the bug hramrach suggested after this
techn_ has quit [Remote host closed the connection]
<wingrime> mnemoc: hm
<wingrime> mnemoc: I rechecked patch again
techn_ has joined #linux-sunxi
<wingrime> mnemoc: but most stuff in #0 was dissabled debug messages or hacks
<wingrime> mnemoc: I can;t find something realy interesing for historic
<mnemoc> removing #if 0 chunks that were only used for debugging is ok. removing #if 0 chunks that provide alternative implementations for usecases we don't know is (imho) wrong
<libv> wingrime: let me take some time to thoroughly review your patches
<mnemoc> removing #if 0 ed hacks that might be critical for certain platforms, doesn't sound good until we really know that case doesn't exist and it was really debugging code
<wingrime> libv: thats mostly hacks
<libv> wingrime: i am perfectly capable of making up my own mind on things
<wingrime> libv: yeax
<mnemoc> *g*
<wingrime> mnemoc: I realy saw sun3i once
<wingrime> mnemoc: but it not cost price for that junk
<wingrime> mnemoc: I now noticed that only one row used from image enchance table
<wingrime> bt709
<wingrime> only
<wingrime> mnemoc: [PATCH] sunxi:disp: expose or remove #ifdef UNUSED codeblocks
<wingrime> mnemoc: There I remove only test colorbars from fb and big unused table
<mnemoc> wingrime: i'm cleaning/packing.... get the ack of another disp-related devs, and if none nacks, the patch gets in AFTER I'm in the next apartment :)
<wingrime> libv: ?
<libv> wingrime: i am busy with FOSDEM at this exact time.
<wingrime> mnemoc: thats all about work finding?
<mnemoc> wingrime: no, it's about having this apartment subrented until tomorrow
<wingrime> libv: thats make my work totaly unproductive
<mnemoc> and have to leave before 10am
<libv> wingrime: you do know about FOSDEM, right?
<libv> wingrime: if not, go read up.
<wingrime> libv: yes, big meetup etc
<libv> wingrime: maybe i shouldn't organize a graphics devroom this year, and put up big signs everywhere "because wingrime his patches couldn't wait"
<wingrime> libv: why you alone do it?
<libv> wingrime: because i manage just fine?
<wingrime> libv: yeax
<wingrime> mnemoc: why so impossible stay on one place? t
<libv> wingrime: just let him live his life.
<libv> it's hard enough as it is.
vinifr has joined #linux-sunxi
<wingrime> libv: I also waste much time hacking that crap code, don't nack me If I want some cleanup before I can do something interesting
<libv> wingrime: you're not making me or mnemoc finish our chores faster by nagging
<libv> and you are not exactly making us keen on rewarding you for this behaviour
<wingrime> libv: I have time now, and don't want lose time to nagging instead of hacking
<libv> wingrime: getting patches accepted takes time.
<wingrime> libv: thats was sended two weeks ago
<libv> wingrime: go do something useful
<libv> wingrime: that's nothing.
<wingrime> libv: example?
<libv> wingrime: oh yeah, let me waste more of my time on digging stuff out just to satisfy your nagging.
<wingrime> libv: why not
<libv> unbelievable.
<wingrime> libv: )
ZetaNeta has quit [Ping timeout: 264 seconds]
ZetaNeta has joined #linux-sunxi
<wingrime> libv: I don't know why I still _have_ so much intension sending it
<libv> 4 minutes.
<libv> a personal best
<tm512> So I've set up a linux from scratch system, has busybox, and a kernel from the linux-sunxi repository, the screen lights up but nothing is printed. shouldn't the kernel at least be panicking if it can't find init or whatever?
<wingrime> tm512: uart
<tm512> what is that?
<wingrime> libv: why every patch _need_ so _much_ polemic
<wingrime> tm512: simple debug interface in our case
<hno> tm512, do you have serial console connected?
<tm512> how do I use that?
<tm512> I have no serial cables to use with the cubieboard
<hno> tm512, get one.
<tm512> conveniently it didn't come with one
<hno> mine did, but might be special order.
<hno> any 3.3V UART cable will do fine.
<tm512> where does it look for init usually?
<hno> /sbin/init or /init depending on how you boot.
<tm512> busybox has it in /sbin
<hno> but there is a million reasons to why something is not printed on screen, and about 2 reasons why someting is not printed on a serial console.
<wingrime> hno: it possible test is on chroot ?
<wingrime> *in chroot
<hno> wingrime, test what?
<wingrime> hno: init
<tm512> oh, yay, more stuff to buy
<hno> no, but you can test it in an lxc container.
<vinifr> :)
<tm512> the only distro I got working with the cubieboard so far was fedora
<tm512> but it has a pretty complicated boot partition
<hno> tm512, anyone intending to do anything more than loading prebuilt Android images really need a serial 3.3V console cable. You will be pretty dead in the water without one.
<wingrime> hno: sometimes uart nothelpfull
<libv> wingrime: example: the sunxi-mali binaries for r3p2-01rel2
<tm512> this is the first time I've done anything resembling embedded systems
<wingrime> hno: I saw coditions after I jump to kernel, I have not saw any output
<libv> wingrime: Benn provided those on august the 13th.
<hno> tm512, how is the Fedora boot partition complicated, apart from having support for many boards and requiring a little basic step to have it configured for the kind of board you have?
<tm512> can I have it dump console to a text file?
<libv> wingrime: did you see Benn whine about them not being integrated in sunxi-mali yet?
<hno> tm512, only if you can manage to get it to run syslogd or the like..
<hno> which is far after init.
<tm512> uboot can't dump to a text file?
<hno> wingrime, I said there is about 2 reasons why something is not printed on a serial console. You hit one of them.
<wingrime> hno: I still not have jtag debuger
<wingrime> libv: I know better example, reizer4
<wingrime> libv: 4 years and still not in mainline ......
<wingrime> libv: btrfs not usable but in ..
<hno> wingrime, bany reasons to why reiserfs4 is not in mainline.
<tm512> to find out if it's my rootfs, I could prepare my system for i686, and then boot it in qemu, right?
<tm512> I dunno if I am missing anything
<wingrime> tm512: try fedora like hno suggessed
<tm512> fedora's boot partition with my rootfs?
<wingrime> tm512: ,,, at least we will know that kernel works
<tm512> ubuntu's boot partiton was simpler and also booted, then I got a plymouth error
<hno> tm512, you can try your rootfs in qemu. Just need a qemu compatible kernel.
<tm512> will mainline work fine?
<hno> tm512, mainline is not something for you at this time.
<tm512> well I already have the source for mainline since I am running it on my computer
<hno> mainline is only meaningful on sunxi if you have a serial console cable, and aware that most drivers are not there.
<tm512> I am just talking about testing for qemu
<tm512> convenient
<hno> sure, mainline armv7 qemu kernel will be fine for testing your rootfs.
<wingrime> tm512: don't forget we need kernel for qemu capable arm maching
<hno> no idea what that link you provided is.
<tm512> it is a download to a kernel that should run under qemu
<hno> tm512, is it an armv7 kernel?
<tm512> but it gives qemu: hardware error: integratorcm_read: Unimplemented offset 0x1f1018
<atsampson> has a fairly handy overview of how to configure a kernel for qemu, so presumably it's the result of that?
<tm512> armv6
<tm512> so nevermind
<tm512> Just remembered that I never installed the modules for my rootfs
<tm512> I'm really on top of things today, as you can see
<hno> you don't really want modules at first testing stage. You need everything needed for fbconsole and accessing your rootfs builtin to the kernel.
<tm512> does the sun4i defconfig give that?
<hno> and then to configure u-boot to boot your kernel with the right console=.. argument.
<tm512> because I didn't mess with it
<tm512> and fedora has a blank boot.scr, does that force it to use uEnv instead?
<libv> wingrime: what did you do with the last 2 hours?
<hno> defconfig quite likely don't give fbconsole. If I am not mistaken the whole lcd/hdmi set of drivers ar all modules.
<libv> wingrime: did you take the advice of techn_ and mnemoc and worked on splitting that one patch?
<tm512> so I should install the modules and then see what's going on
<hno> tm512, u-boot reads both, first uEnv and then boot.scr if uEnv haven't told anything else.
wingrime has quit [Ping timeout: 256 seconds]
leviathanch_ has quit [Remote host closed the connection]
leviathanch has joined #linux-sunxi
<tm512> gonna try with the modules actually installed this time
<tm512> still nothing
<tm512> cubie has both lights on
<Turl> mripard: ping
<tm512> my bootargs are console=ttyS0,115200 disp.screen0_output_mode=1920x1080p60 root=/dev/mmcblk0p2 rootwait panic=10
ZetaNeta has quit [Ping timeout: 264 seconds]
ZetaNeta has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi