<arokux2>
Turl, I've discovered something. 3.4 code used IRQF_SHARED | IRQF_DISABLED. in the mainline IRQF_DISABLED seems to be deprecated and said to be noop, which seems to be the case.. however eliminating IRQF_DISABLED did remove the eacterr things, but wifi still won't work.
<arokux2>
Turl, do you know something about IRQF_*?
<Turl>
arokux2: nope :(
<Turl>
arokux2: maybe there is a bug on your irq handling?
<Turl>
arokux2: do you have a link handy to your patches?
<arokux2>
this is aw code, it enables the clock, doesn't it?
kaspter has joined #linux-sunxi
<arokux2>
then there is clk_reset(sw_hci->phy_reset, 0); -- which is noop (at least on sun4i). and after commenting it out in aw code wifi still works.
<balage>
arokux1, no I am diffing the kernel config of 3.0 and 3.4 to see what did I miss
<oliv3r>
balage: so 3.0 works; 3.4 doesn't work?
<oliv3r>
balage: that's quite an important change :)
<oliv3r>
balage: leviathanch usb_host_init_state enables/disables default power
<oliv3r>
so with initstate = 0; usb is unpowerd, and wifi won't work
popolon has joined #linux-sunxi
Xaros has quit [Quit: Zwi se logou mou!!!]
Black_Horseman has joined #linux-sunxi
<balage>
oliv3r, there are three usbc parts in fex file, usbc0 usbc1 and usbc2. usb_host_init_state has been set to 1 for usbc0 and usbc1 but set to 0 for usbc2. is that a problem?
<balage>
oliv3r, forget it, i watched the old fex file from backup and not from github
<balage>
oliv3r, so in the latest fex all three has been set to 1
HeHoPMaJIeH has quit [Ping timeout: 260 seconds]
<balage>
oliv3r, I diffed the usb parts from config 3.0 and 3.4, would you please take a look? http://sprunge.us/aJAH
HeHoPMaJIeH has joined #linux-sunxi
<JohnDoe_71Rus>
green robot was 5 years old
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
focus_it has quit [Remote host closed the connection]
shineworld has quit [Remote host closed the connection]
atiti has quit [Ping timeout: 260 seconds]
hansg has quit [Ping timeout: 264 seconds]
<Tsvetan>
oliv3r ping
<oliv3r>
Tsvetan: pong
focus has joined #linux-sunxi
<oliv3r>
balage: mUSB is usb0 by default; so host0 and host1 are both disabled; e.g. no power for your wifi. it is safe to have all 3 set to 1; actually for now i recommended it
<oliv3r>
recommend it*
<oliv3r>
balage: -CONFIG_USB_SUPPORT=y ??
<oliv3r>
balage: oh nvm it moved :)
<oliv3r>
balage: a sucky diff
<oliv3r>
balage: looks reasonable
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
<oliv3r>
balage: i'm trying to boot my own tablet again tonight with 3.4; i'll let you know then
<balage>
oliv3r, thanks!
<oliv3r>
balage: just dont' see what you might be missing or have done :p
atiti has joined #linux-sunxi
atiti has quit [Ping timeout: 260 seconds]
_enrico_ has joined #linux-sunxi
FDCX has quit [Ping timeout: 252 seconds]
<mnemoc>
OT: what was the name of amd's? arm board with itx-ish form factor?
<oliv3r>
erm
<oliv3r>
there was such a thing?
<n01>
here @ work I have a NUC ... cool thing
Guest93817 is now known as eebrah_
eebrah_ is now known as eebrah|FTS
hipboi_ is now known as hipboi
<mnemoc>
oliv3r: yes, a very stupid armv11 iirc
<oliv3r>
sounds old
<mnemoc>
but itx or atx case compatible
<mnemoc>
it was announced last year, and recently got an upgrade
<mnemoc>
at least their latest model is cortex-a9, but 512MB :p
<oliv3r>
case looks nice
<oliv3r>
well if it's not shared with the GPU
<oliv3r>
it might be 'okay'
<oliv3r>
-ish
<mnemoc>
Neo-ITX is form factor compatible with Mini-ITX and MicroATX
<oliv3r>
looks like it has a PCI bus
<mnemoc>
i only wanted an example pc-ish board for hipboi
<rm>
still not "amd's"
<oliv3r>
ahh ok neo-itx is a really nice formfactor
<mnemoc>
rm: that's why I said "amd's?"
<oliv3r>
it'll fit into many itx cases
<mnemoc>
exactly my point
<rm>
oh right, I am not into proper usage of mid-sentence question marks :)
<mnemoc>
rm: :p
<oliv3r>
Tsvetan: neo-itx boards :D
<oliv3r>
brb
<mnemoc>
rm: VIA :)
<rm>
I like the image they try to create around it
<rm>
but doooon't like that they push Android as the OS
akaizen has joined #linux-sunxi
tinti has quit [Ping timeout: 252 seconds]
tinti has joined #linux-sunxi
soldoKyn has joined #linux-sunxi
hipboi has quit [Quit: Leaving]
<arokux1>
balage, have you got it working?
<arokux1>
does somebody have a wireless usb dongle?
<JohnDoe_71Rus>
yes
<arokux1>
JohnDoe_71Rus, what sunxi board do you have?
<JohnDoe_71Rus>
cubieboard2
<arokux1>
JohnDoe_71Rus, no UART2USB?
<JohnDoe_71Rus>
arokux1: No. I have not tried Linux, I have no usb keyboard
<oliv3r>
Turl: feel free to rm -r dl.linux-sunxi.org/users/turl/oliver/ it's old and don't even know what exatly it contains :)
<arokux1>
oliv3r, (the current git is broken now )
<arokux1>
JohnDoe_71Rus, perfect. maybe you can help me with testing? have you booted mainline already?
<arokux1>
oliv3r, you've promised to test my drivers :p
<oliv3r>
arokux1: yes i have one
<oliv3r>
arokux1: this week
<arokux1>
JohnDoe_71Rus, i.e. do you know how to etc.
<arokux1>
JohnDoe_71Rus, do you run X on it?
<oliv3r>
arokux1: ok when you have a git tree that I can test; let me know :)
<arokux1>
oliv3r, !! I had it. now you say you wait for me, I'll have it tonight. will you test tonight? :p
<oliv3r>
if you help me not forget :(
<oliv3r>
but i have to test 3 things anyway tonight
<oliv3r>
so busy busy busy :)
<arokux1>
oliv3r, I'll remember you each hour, no worries
<arokux1>
fucking Chinese code is ugly but works :(
<JohnDoe_71Rus>
arokux1: i test only android. But I do not have drivers for android. For the experiments, first i need to get the boot sd card
<arokux1>
JohnDoe_71Rus, ok, I see. too much to do for you. (you can do if you want, though) oliv3r has a wifi dongle, and he's familiar with mainline
tinti has quit [Ping timeout: 260 seconds]
<arokux1>
JohnDoe_71Rus, so he can test.
<oliv3r>
arokux1: i have the same :p though I haven't investigated why it doesn't work :)
<oliv3r>
just ianc reported it doesn't work :)
<arokux1>
oliv3r, well.. I didn't see you investing enough time into it :p
<arokux1>
oliv3r, but *once* it works for you --- I will test immediately.
<oliv3r>
how can you see me! :(
<oliv3r>
you got cameras?
<JohnDoe_71Rus>
big brother is wathing you )
<oliv3r>
are you in my house?
<arokux1>
my advice to you guys. NEVER EVER optimize code while porting. port it ugly as it is, make sure it works and only then step by step improve it.
<oliv3r>
arokux1: of course :D
<arokux1>
I wish there were something like memory isolation layer, so I can see what chinese code really flushes to the memory etc.
<atsampson>
arokux1: first two rules of optimisation: (1) Don't. (2) Don't *yet*.
<arokux1>
but the func I used additionally writes to memory.
<DigiDaz_>
Hi all, I have a cubieboard2. From my very limited knowledge, what I seem to be seeing is that the debian armhf builds are noe optimized for a particular processor would I see much performance improvement by using say a gentoo build?
<buZz>
no
<ssvb>
DigiDaz_: it depends, for mplayer/ffmpeg you will see a major performance improvement after enabling neon, for the other software - not so much
<DigiDaz_>
Thanks for that, its actually for little PBX boxes I'm creating, either asterisk or freeswitch so I guesss it would be time largely wasted then. They will be headless.
<oliv3r>
arokux1: are you sue they are not writing it?
<oliv3r>
i see, temp = readbyte; change Change BIT(7) in temp; write temp to dest
granite_crusher2 has joined #linux-sunxi
<oliv3r>
and bit 7 is either cleared or set depending on dtmp &0x1 state
<oliv3r>
arokux1: so you are doing completly different thing there, omitting BIT(7)
<granite_crusher2>
arokux1: and don't forget I can test on A13, and I have some USB stuff (pendrives, few mouses, few keyboards, USB1.1 pro sound card, USB ethernet dongle..) :)
<granite_crusher2>
arokux1: if to put more in the list: USB MIDI keyboard
<oliv3r>
granite_crusher2: intersting is if usb 1.1 devices work; in theory, it shouldn't. i'm curious i it does, would be cool if it did
<granite_crusher2>
arokux1: are usb drivers missing just usb part in DT in mainline?
<granite_crusher2>
arokux1: I mean for A13
\\Mr_C\\ has quit []
gnufan has quit [Ping timeout: 256 seconds]
<arokux1>
oliv3r, no, not completely different, but I write two times - this is a problem.
<arokux1>
granite_crusher2, mostly DT bits.
\\Mr_C\\ has joined #linux-sunxi
<oliv3r>
arokux1: 'sort of different' then ;)
<arokux1>
oliv3r, writing a bit to memory triggers some hardware event, I suppose..
<oliv3r>
well they are doing writeb, so it may even trigger 32 bit alignement issues ontop of that
<arokux1>
oliv3r, rw same content should do any difference to hardware?
<arokux1>
interesting thing is if this UsbPhyInit is omitted, the first controller won't work, but the second one still will.
Black_Horseman has joined #linux-sunxi
hno has quit [Ping timeout: 260 seconds]
<oliv3r>
arokux1: the hardware can respond quite differently, while I don' think it's likly; yes, its quite possible the chip will do something the moment you write the value to it
<oliv3r>
so changing write order in that sense, is probably not wise
<oliv3r>
unless you have documents that tell you what it exaclty does (which we don't have right now)
<arokux1>
only one think bothers me. aw's code work - mine doesn't.
<oliv3r>
example: BIT(31) = send BIT(0-7).
<oliv3r>
so if you first do BIT(31), and then fill BIT(0-7) you send 00000 :)
hno has joined #linux-sunxi
<oliv3r>
so without knowing exactly what certain bits do; you can only guess and changing order/split writes, bad things will happen (tm)
<tgaz>
the only times i had an issue with my a20 was when i forgot to include the hdmi driver in the kernel. is it just X not working, or is FB also broken?
<orsic>
hmm
<ssvb>
orsic: dmesg output looks incomplete, for example there is no information about the kernel cmdline
<oliv3r>
that's my wip thing, it'll need to be split, moved, fixed
<steev>
i have a monitor here that's the same, very few devices can handle the pixclk at 1680x1050
<oliv3r>
but it 'works' so i haven't spent much time on it
<orsic>
steev: hmm thats unfortunate
<oliv3r>
mnemoc: the bsp does need some love, a couple of hours I suppose; i just don't wanna yet :p more important things to do :)
<steev>
orsic: not a whole lot can be done there. pixclk is pixclck
<oliv3r>
mnemoc: oh shit, i pushed my partprobe patch that you nacked ages ago
<orsic>
right
<oliv3r>
i still need it :p
<orsic>
is there somewhere a table of suported pixclk's?
<oliv3r>
oh the patches are messy :) i should clean them up and rename it etc etc
<oliv3r>
mnemoc: btw, so far, it works for all 4 kernels, mailnine, 3.10, 3.4 and 3.0
<mnemoc>
\o/
<mnemoc>
i think we should try to make the bsp make a multisunxi bin, without embedding the dtb
<mnemoc>
but haven't really looked at your branch yet
<mnemoc>
making easier to provide nighly builds, and encouraging testings
<mnemoc>
also teaching fexc to render .dts files
<mnemoc>
out of the .fex
<oliv3r>
oh that's all early experiments
<oliv3r>
that's really old commit; i did that because I wanted to append the dtb without messing with u-boot etc
<oliv3r>
it can be completly dropped now actually, it's useless
<mnemoc>
:)
<oliv3r>
as I said, it's old wip stuff, that i still use and haven't bothere dcleaning up :)
<mnemoc>
what about "next" instead of "mainline_support" ?
<oliv3r>
i'll rename the branch anyway and delete the one pasted
<mnemoc>
ok
<oliv3r>
and I should have probably done a pull/merge :)
<oliv3r>
but yes, it needs som love
<mnemoc>
nah, rebase.... merge commits suck
balage has left #linux-sunxi [#linux-sunxi]
<oliv3r>
yeah rebase; but it won't let me
<mnemoc>
o_O
<oliv3r>
the branches have diverged
<mnemoc>
so?
<oliv3r>
so i can't rebase bla bla
<oliv3r>
git stash i guess
<mnemoc>
you can resolve conflicts too ;-)
<arokux1>
mnemoc, do you know something about dts? it is something completely different from fex.
<mnemoc>
arokux1: yes, but the .fex tells what hw is in the board, and the dts needs that info
<n01>
mnemoc: stupid question: does exist a fb driver for cb?
<oliv3r>
mnemoc: yeah but it says unmerged stuff on the branches
<oliv3r>
which i don't care about :p
<mnemoc>
cb1, yes. cb2... no idea
<mnemoc>
oliv3r: git reset --hard
<oliv3r>
i'm not trashin' my subtree's!
<oliv3r>
submodules!
<arokux1>
mnemoc, dts needs way more than that. i doubt you can write a fex2dts at all, doing it by hand is faster.
<n01>
oliv3r: you know anything about a fb driver for cb/cb2?
<oliv3r>
i manually went into submodules,, did git remote add; git push/pull etc
<oliv3r>
n01: never used it :)
<oliv3r>
n01: well knowingly :p
<n01>
uhm so not mainlined
<oliv3r>
arokux1: for each and every board?
<oliv3r>
n01: oh hell no
<oliv3r>
n01: i wish :D
<mnemoc>
arokux1: the resulting dts will obviously start with the per-soc import. by hand for every single sunxi-based device.... doesn't scale
<n01>
hum. ok @ work I have to write a new fb driver a I could use cb to make practice
<mnemoc>
arokux1: and the dts model will evolve with time.
<mnemoc>
arokux1: easier to maintain the .fex, and compile the .dts
<arokux1>
mnemoc, ok, I encourage you to take a look at the dts, you'll understand dts and fex are different worlds.
<arokux1>
mnemoc, dts based devices need IO_MEM, clocks.. nothing like this in fex.
<oliv3r>
once we gotten rid of fex at all; dts all the way
<oliv3r>
arokux1: they are not so different
tzafrir has joined #linux-sunxi
<oliv3r>
arokux1: those mems, clocks are all defined in the dtsi*
<oliv3r>
*.dtsi
<oliv3r>
the dtsi you write by hand, the dts you can somewhat automate
<mnemoc>
all that stuff comes from the handcrafted per-soc dti
<arokux1>
well.. then there is not much in dts you can grab from fex
<arokux1>
only GPIO pins
<mnemoc>
the per-board part, can be done out of the .fex and some static strings in the .c
<arokux1>
and status=disable/okay (what an idiot selected them like this? why not on/off disable/enable good/bad...???)
<mnemoc>
arokux1: even if only the muxing, teaching the fexc to do so is easier, user friendlier and more reliable than manually writting a .dts for each sunxi board
<mnemoc>
and until AW joins us, new sunxi boards will keep coming with script.bin files
<mnemoc>
which fexc can perfectly turn into myboard.dts
<mnemoc>
assuming there is already a per-soc .dti
<arokux1>
ok, if you write dti per hand then something is possible
<arokux1>
you'd need however to parse dtsi too for converting from fex too dts
<mnemoc>
initially I would just hardcode some hints per-soc in the module in charge of writting the dts
<arokux1>
lots of work. and no drivers to take advantage of these autogenerated dtses
<arokux1>
maybe drivers should come first? :p
<mnemoc>
the point of 3.10 is to DTSize crap drivers
<mnemoc>
and get a real-world usable DTS-based sunxi kernel
<mnemoc>
and use that for real-world testing and for making crap drivers reach mainlineability qualitz
<mnemoc>
quiality
<mnemoc>
an step stone
<mnemoc>
but usable in real life
<mnemoc>
all drivers getting into 3.10 will need .dts data
<mnemoc>
at least to know the muxing and what periferals are actually there
<arokux1>
maybe 3.4 should have been selected as stepping stone :(((
<arokux1>
I'm not seriously saying this
popolon has quit [Remote host closed the connection]
<mnemoc>
3.4 is for unification and aw-source-drops mostly
<arokux1>
just thinking why my usb still won't work fully....
<mnemoc>
changes that need to be addresed to reach 3.13 anyway
<mnemoc>
once a 3.10 driver becomes not-shameful, mainlining can begin
<arokux1>
mnemoc, yes, this is the problem that I see
<mnemoc>
there are more changes affecting us between 3.4 and 3.10 than between 3.10 and 3.13
<mnemoc>
but in 3.10 they can be tested
<mnemoc>
once unified, next step to DTSize in 3.10
<arokux1>
mnemoc, I see. but many things changed in 3.10, and if something won't work there you do not know what the problem is
<mnemoc>
3.10 is an stable stepping stone. not work duplication
<arokux1>
i'm not advocating anything, i'm just saying it is difficult to port to the highly change environment :(
<mnemoc>
we need to port drivers from 3.4 to master anyway
<arokux1>
my usb glue still works only partially and I do not know why..
<mnemoc>
easier to do so to a 3.10 that gets TESTED
<mnemoc>
than to a master barely tested by some devs
<mnemoc>
once it works well in the stable 3.10, the little changes to reach master can be done reliably
<arokux1>
mnemoc, I can test my drivers by myself.. my tests are enough to tell if the are 90% working.
<mnemoc>
an stable brach with nightly builds can push that to a 99%
<arokux1>
mnemoc, the only problem was different boards: but soon I'll have three: A10, A20 and A13
<mnemoc>
because of real life tests in assorted real hardware
<mnemoc>
3.10 solves that
<mnemoc>
it's not merely about the SoC. different PCBs might need different hacks which your driver is not considering
<mnemoc>
and people using .310 will notice
<mnemoc>
you won't
<arokux1>
mnemoc, I agree. however my focus is a bit different. I won't to learn Linux in the first place.
<mnemoc>
s/won't/want/
<mnemoc>
3.10 will teach you linux too
<mnemoc>
refactoring crappy code into linux quality will teach you even more
<mnemoc>
and can give you far more testers than lkml
<mnemoc>
specially if we move ALL crap drivers
<mnemoc>
so the thing actually becomes usable
<arokux1>
mnemoc, nobody will give you an advice if you ask about the code, however, if it is not mainline
<mnemoc>
arokux1: there are several professional kernel developers in the sunxi list too
<arokux1>
mnemoc, haven't seen them yet.. except of mripard nobody is helping.
<mnemoc>
mripard is not nobody
<mnemoc>
neither is hansg or steev
<mnemoc>
and the first step into 3.10 is only to DTSize
<steev>
wat
<arokux1>
mnemoc, reread. "except of mripard nobody is helping"
<arokux1>
mnemoc, DTS is not a big deal!!
<mnemoc>
< arokux1> mnemoc, nobody will give you an advice
<arokux1>
DTS is the smallest problem you can have
<steev>
i'm not a professional kernel developer. i do kernel development, but i'm not professional
<arokux1>
it is painless.
<mnemoc>
the jump from 3.4 to 3.10 is not painless, but it's only about code changes
<mnemoc>
the jump to master is more painless, includes everything you already did for 3.10, and style control
<arokux1>
mnemoc, not dts is the problem, but the clk framework and using proper kernel frameworks. dts is *very* easy.
<mnemoc>
therefore 3.10 is an stable middle ground
<mnemoc>
steev: are you paid for belly dancing or kernel development?
<arokux1>
mnemoc, what you say has sense, you shouldn't repeat it so many times (unless you've got paid to do so :p). it just more work, that is all.
<WarheadsSE>
steev: twat
<mnemoc>
arokux1: I'm paid to maintain/refactor/fix chinese proprietary crap code :(
<mnemoc>
arokux1: i'm trying to explain that it's not really more work. it's to do the same job, in an stable branch, all together instead of in random personal repos
<mnemoc>
and migrating good 3.10 drivers to master is not that much work. most will be already done
<WarheadsSE>
mnemoc: ever seen the ZyXEL crap?
<WarheadsSE>
Trying to figure that watchdog wiring bullshit out to merge into 3.11+
<WarheadsSE>
>.<
<WarheadsSE>
they attached a MCU with access to the power gpio trigger, and you can only communicate with it over the LED[2] of the ethernet phys, and only if you have set said LED to tristate via register manipulation
<WarheadsSE>
ANd if you don't make it happy, it slowly reboots the box every X seconds on power up with increasing time until it works. ...
<arokux1>
WarheadsSE, why 3.11+?
<arokux1>
(and not 3.12+)
<WarheadsSE>
arokux1: because it is a kirkwood based system
<WarheadsSE>
and I am a core developer at Arch Linux ARM
<arokux1>
WarheadsSE, yeah, I know.
<arokux1>
and kirkwood base kernel is 3.11 or what?
<WarheadsSE>
3.12+ in regards to this ethernet phys is identical to 3.11+
<WarheadsSE>
the kirkwood was mainlined long ago
wingrime has quit [Ping timeout: 260 seconds]
<arokux1>
ok. btw, WarheadsSE do you know smth about SATA port multipliers i.e. if there are kernel drivers for some decent ones..?
wingrime has joined #linux-sunxi
<WarheadsSE>
smth I assume to be "something"
<arokux1>
WarheadsSE, yep
<WarheadsSE>
the host port has to support it
<WarheadsSE>
otherwise, it won't matter
<arokux1>
WarheadsSE, do you know if sunxi SoCs' host port supports that?
<WarheadsSE>
All sheeva-based cores (kirkwood, mvebu, dove) support port mulitipliers
<WarheadsSE>
imx's tend to not
<WarheadsSE>
I can't say I have tried it
<arokux1>
WarheadsSE, how can you try, do you have hardware for multiplication?
<WarheadsSE>
it should, in theory, baring hardware faults of logic, be possible to add it
<WarheadsSE>
I have a BlacX 2 port multiplier, yes
<arokux1>
WarheadsSE, you know, it is just interesting if sunxi can do some decent NAS
<WarheadsSE>
Yes.. I know what you are asking
<WarheadsSE>
but even in them, it states that the host port has to have multiplier functionality
<WarheadsSE>
I have not looked into if it does
<WarheadsSE>
(and don't have time right now)
<arokux1>
WarheadsSE, no problem. we do not have drivers in mainline yet too. maybe you can do once you have time.
<WarheadsSE>
bahaha
<WarheadsSE>
yeah... that's why I haven't done much witht he sunxi kernels yet
<arokux1>
WarheadsSE, you've planned your time till death? :)
<WarheadsSE>
they are so heavily under development, and I don't have a lot of free time
<arokux1>
WarheadsSE, sunxi-3.4 is feature full, though
<arokux1>
yes, regressions happen.
<mnemoc>
*hint* 3.10 *hint*
<mnemoc>
:)
* mnemoc
is very annoying at advocating
<arokux1>
mnemoc, what? there will be regressions too.
<arokux1>
mnemoc, best way to advocate is to backport from mainline to 3.10, the others will follow!
<mnemoc>
but been feature complete more people will use it, and more people will fix it
brain__ has quit [Quit: Leaving]
brain__ has joined #linux-sunxi
<mnemoc>
mainline has a quality restriction 3.10 doesn't
<akaizen>
Is Allwinner A20 capable of HDMI-CEC or is that board specific?
<mnemoc>
didn't drachensun get libhybris working on A31's PVR?
<arokux1>
mnemoc, with A31 being powervr and the community disguising it we are loosing some user base though.
<brain__>
ok thanks mnemoc
<arokux1>
mnemoc, lib
<mnemoc>
brain__: /nick brain___
brain__ is now known as brain_
<brain__>
how do i change my nickname so i can join a channel were i need authentication? i think their was another _ added sometime
<arokux1>
libhybris is just a wrapper though.
<mnemoc>
better than nothing
<drachensun>
a wrapper that makes it work
<drachensun>
60 fps :-)
<mnemoc>
and let's you run X on A31 nicely
<arokux1>
mnemoc, i'm saying the same. there is a need for headless systems too. i do not care for GPU unless for computations.
Soru has quit [Read error: Connection reset by peer]
<mnemoc>
sure
<arokux1>
but certain ppl here think A31 shouldn't be looked at *at all*, because of powervr
<mnemoc>
drachensun: want to move your libhybris thing to the sunxi repo?
soul has joined #linux-sunxi
<mnemoc>
drachensun: and btw, did the UPS parcel arrive?
<drachensun>
mnemoc: by the way, UPS did manage to deliver the devices last Thursday, no drama about hazardous anything, wish I had just used them in the first place
<mnemoc>
or that box is still in limbo?
<brain_>
i hate powervr due to how much of a pain it is to get the blob working on my beagleboard. such a hassle
<mnemoc>
drachensun: good to know for next time
<drachensun>
yup
<drachensun>
mnemoc: so yeah, we are back on track, press samples went out
<mnemoc>
nice
<drachensun>
hopefully we will get some press again and get the funding moving
<drachensun>
the libhybris thing was just a build of latest version from git, on the device running mer
<drachensun>
I didn't have to modify it any
<drachensun>
I did have to have to put all the powervr libraries and setup files in the right place
<WarheadsSE>
drachensun: could you explain the *actual* requirements of libhybris?
<drachensun>
and make a few sym links for library names
<mnemoc>
drachensun: no bindings?
<drachensun>
well basically it all follows pretty well from vgrades guide
<WarheadsSE>
I keep getting descriptions that basically leave me needing an entire android chroot
<drachensun>
warheadsSE: Well, the entire /system more or less, you can skip the directory with all the apks
<drachensun>
but yeah /system and /vendor from android
<drachensun>
put in the right place under Linux, either with a bind
<drachensun>
or if my case I just put them there on root
<WarheadsSE>
blech
<WarheadsSE>
how big of a crap dump of disk space is that
<drachensun>
it probably could be trimmed down
<drachensun>
300 mb?
<WarheadsSE>
that isn't *horrible*
<drachensun>
something like that,
<drachensun>
most of the junk in /system is all the android apks
<WarheadsSE>
If it could be trimmed down the the bare requirements, I would package it for the various boards ALARM works with
<drachensun>
and there was one other big folder we trimmed off, I don't recall exactly
<WarheadsSE>
but there is no way I am making that a package, with one of the methods I saw (cubie) having essentially a complete 2G android image ...
<WarheadsSE>
because FFS, no way
<drachensun>
I guess one could get it running
<drachensun>
and then pull strace to see what libraries are actually used
<arokux1>
WarheadsSE, are you sure the host SATA port hardware need to have some specific support in order for the multiplier to work?
<WarheadsSE>
arokux1: nope, I talk out my ass all the time
<arokux1>
WarheadsSE, sorry, I'm just double checking, got some other opinion in #arm-netbook
<WarheadsSE>
oy
<arokux1>
WarheadsSE, ppl will try to build a NAS around sunxi
<WarheadsSE>
yes.. and they can
<WarheadsSE>
but you need to hook it up and try, then debug if it doesnt work
shineworld has quit [Ping timeout: 248 seconds]
ykchavan has joined #linux-sunxi
soldoKyn has quit [Quit: Konversation terminated!]
<brain_>
has anyone tried using their mmc port for JTAG debugging? I have Segger Jlink and another knockoff i could use to see what happening with my touchpanel module
<arokux1>
"I guess you need some more "meat" in your commit log." :)
Soru has joined #linux-sunxi
<Turl>
brain_: pretty sure hno did, maybe oliv3r too
<Turl>
arokux1: USB isn't that bad for a NAS which has 100Mbit eth
<arokux1>
Turl, true. 100Mbit is too little though.
<Turl>
arokux1: you can make the case again when CT ships :p
<arokux1>
Turl, again?
<Turl>
I'm just kidding :)
<Turl>
CT has GbE phy iirc
<arokux1>
Turl, :) yep. GMAC.
<arokux1>
(plus 1Gbit PHY, I suppose)
ganbold_ has quit [Remote host closed the connection]
popolon has quit [Quit: Quitte]
ykchavan has quit [Ping timeout: 252 seconds]
_enrico_ has quit [Quit: Bye]
FDCX has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
wingrime has quit [Ping timeout: 268 seconds]
wingrime has joined #linux-sunxi
arokux2 has joined #linux-sunxi
Seppoz has quit [Ping timeout: 260 seconds]
<brain_>
WOOT! Fixed Netflix! working better than stock firmware were it would freeze after a minute
<oliv3r>
brain_: grats :)
<brain_>
Thanks, working on the camera now..
_whitelogger___ has joined #linux-sunxi
<arokux2>
oliv3r, I just do not have usb wifi dongle :((
<oliv3r>
arokux2: i'll test it soon :)
<arokux2>
oliv3r, this week you said.
<oliv3r>
first 3.4 on tablet; then sata ml driver; then usb ml
<oliv3r>
you have git tree?
wingrime has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
FR^2 has quit [Quit: Connection reset by peer]
massi has quit [Remote host closed the connection]
<arokux2>
oliv3r, you said this week, so no matter what it should be this week ^^
<arokux2>
oliv3r, who you think will test SATA for you?;)
<oliv3r>
iac did
<arokux2>
oliv3r, and I *will*
<arokux2>
oliv3r, do you want me to provide base config for you? (you'd need to add drivers for you usb devices by yourself)
wingrime has quit [Ping timeout: 245 seconds]
deasy has joined #linux-sunxi
wingrime has joined #linux-sunxi
pitillo has quit [Changing host]
pitillo has joined #linux-sunxi
<oliv3r>
sorry, mommy on the phone :)
jemk has joined #linux-sunxi
<oliv3r>
hno: i'm using the lichee-dev nand bootloader and try to load a uImage (latest 3.4) that i've copied to /dev/block/nanda. I use the regular bootm dance, but get the machine ID thing I have seen pass by here. What's causing this? http://sprunge.us/
<oliv3r>
i purposly skipped the initramfs bit
<Turl>
oliv3r: your sprunge url is cut
<oliv3r>
crap
<arokux2>
granite_crusher, hey, I've pused support for A13 and A13-Olinuxino. want to try it out?
<oliv3r>
Turl: i'm not all up to speed with the old :p
<oliv3r>
Turl: that's sun7i
<oliv3r>
Turl: my tablet is sun4i
<Turl>
ah
<granite_crusher>
arokux2: ok I am going to try
<Turl>
oliv3r: yeah then 1008
<oliv3r>
for the logs: sun4i#set machid 0x1008
<oliv3r>
Turl: boots :D
<Turl>
oliv3r: set works? I always env set or setenv :p
<oliv3r>
:)
<Turl>
oliv3r: great :)
<oliv3r>
so now to try with initramfs
<oliv3r>
as it rebooted at the end
<oliv3r>
No filesystem could mount root, tried: ext3 ext2 ext4 vfat msdos iso9660 ntfs fuseblk udf
<arokux2>
granite_crusher, next (after make multiv7.... ) run make menuconfig and go to Device Drivers/USB Drivers and select EHCI HCD (USB 2.0) support and then a bit lower: Allwinner sunXi EHCI support
<oliv3r>
hopefully initramfs fixes that
<arokux2>
granite_crusher, next, you'd need to add drivers for you USB dongles.. if you do not know their names stick them into your desktop, and check dmesg | tail, find the name of the driver and do: git grep <DRIVER_NAME>. this should give you a clue what config option to enable (this can be a bit tricky..)
akaizen has joined #linux-sunxi
<oliv3r>
Turl: what's a safe address to load the initramfs when kernel is at 0x48 and script.bin is at 0x43?
<arokux2>
granite_crusher, you can search in menuconfig pressing "/", you should input config options there. search results should give you a clue where the specific driver config is buried and what should be (possibly) enabled to even see him. normally drivers depend on some features you need to enable first.
<arokux2>
granite_crusher, I'm leaving now for 2 hours or so, hope everything works!
<Turl>
oliv3r: 0x50...?
<oliv3r>
kernel is about 4mb so should easily fit
<oliv3r>
Ramdisk image is corrupt or invalid
<oliv3r>
*sigh*
<Turl>
oliv3r: make sure you have the ram
<arokux2>
oliv3r, was it *.cpio?
<oliv3r>
512 mb
<oliv3r>
nah .gz :)
<Turl>
oliv3r: I don't recall if 0x40->0x50 was 256 or 512
<arokux2>
gz didn't worked for me... (athough shoud?)
<oliv3r>
i'll unzip the gz and use the raw cpio
<Turl>
arokux2: I'll see if I can test that usb soundcard
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<oliv3r>
Wrong Ramdisk Image Format
<oliv3r>
that's the cpio :S
<oliv3r>
the cpio I friggin' extracted from the original boot.img
<oliv3r>
initramdisk: ASCII cpio archive (SVR4 with no CRC)
<oliv3r>
by any chance lichee-dev u-boot only support boot.img?
<oliv3r>
Turl: anyway, the machine ID could be the primary reason why my board kept failing to boot! so i'll hack da7 into the kernel (i'd rather leave u-boot as-is)
<Turl>
oliv3r: did you uImage it?
<Turl>
uboot loves uImages
<oliv3r>
yeah
<oliv3r>
but the ramdisk failed
<oliv3r>
thew kernel is happy uImage
<oliv3r>
i'll build a 3.4 including initramfs to make it easier :D
<oliv3r>
bfree: i extracted the initramfs from the boot.img
<oliv3r>
so it should be sound
<oliv3r>
but of course, i get no debug output from my kernel anymore either *sigh*
<oliv3r>
time for usb kernel
<oliv3r>
Turl: run what? i boot kernel from console
<oliv3r>
Turl: nvm i was reading old log :S
<bfree>
oliv3r: link has u-boot sample code for 3.4 and mainline, mkimage command, gotcha ("setenv initrd_high 0xffffffff" which I needed anyway for mainline on cubie)
Tartarus_ has joined #linux-sunxi
Tartarus_ has quit [Read error: Connection reset by peer]
<oliv3r>
bfree: i thought initrd_high was needed for really big initramfs's due to relocating going worng
<oliv3r>
bfree: anyway, i was trying to liberate my tablet; which still runs a gpl violating kernel; boot.img however never booted, no output, nothing
<oliv3r>
bfree: so i extracted boot.img, split it into initramfs and the kernel, copied my own 3.4 kernel + the initramfs to nanda and booted it via u-boot console
<oliv3r>
bfree without initramfs the kernel boots fine, but craps out due to missing root, understandable, but it won't eat my initramfs
<oliv3r>
bfree: now i'm guessing, lichee-dev u-boot doesn't support initramfs :p
<bfree>
afaik initrd_high is needed for mainline (dtb)
<bfree>
hmm, I don't know how to un-mkimage an initrd to really check that initrd (or remake it) :-/
<bfree>
and also have no idea about lichee shite :-p
<oliv3r>
bfree: yeah but if i want to boot from nand; i'm kinda stuck with it ;)
<rz2k>
linaros gcc 4.9 is out
<bfree>
tried mkimage on it? maybe it's an un-mkimaged initrd
<oliv3r>
Turl: since when do we have 'physical base address of debug UART' just below the DEBUG_LL? and do we actually have to enter it?
<WarheadsSE>
rz2k: now to see if it doesn't break like hell
<Turl>
oliv3r: dunno
<rz2k>
WarheadsSE: compiling for sun7i right now
<rz2k>
gonna test both kernel and u-boot
<oliv3r>
Turl: :(
<oliv3r>
arokux2: your kernel doesn't output any debug info either :S
<WarheadsSE>
oliv3r: just set the loglevel=8 ?
<Turl>
arokux2: no luck
<Turl>
arokux2: I think it may be OHCI
<Turl>
arokux2: can you implement it? :)
<oliv3r>
arokux1: i was affraid of that
<oliv3r>
maybe my kernel is just crashing
<oliv3r>
Turl: multi_v7_defconfig doesn't do debugging by default does it?
<Turl>
oliv3r: _LL? no
<oliv3r>
ok i'll try 3.10 with the sunxi_defconfig
<Turl>
arokux2: once I plug that device the port stops working, no more interrupts when plugging good devices
<Turl>
it may be switching into ohci mode and as there's no driver it stays there or sth
<oliv3r>
i think doing OHCI first; and then EHCI would have been a better combination
<oliv3r>
then again, EHCI should work
<oliv3r>
but I think i noticed the exact same behaviour on x86
hugobo has joined #linux-sunxi
<oliv3r>
Turl: what defconfig do you use
<oliv3r>
multi_v6?
<hugobo>
I have a question related to boot.axf: I tried to replace boot.axf with another ELF executable, but I now have the error message "elf file header magic not match" when trying to boot. How is that "elf file header magic" comparison done?
FDCX has quit [Ping timeout: 268 seconds]
massi has joined #linux-sunxi
<Turl>
oliv3r: v6?
<oliv3r>
v7 :p
<Turl>
oliv3r: I use something derived from v7, or sunxi_defconfig on my repo
<oliv3r>
so far, nothing i tried has worked; 3.4 is half a failure; next-usb isn't printing debug; ahci isn't printing debug
<oliv3r>
only u-boot prints its stuff luckly
<Turl>
hugobo: it's hard to know when the code is proprietary :(
<Turl>
oliv3r: did you enable DEBUG_LL, choose the right sunxi uart, enable earlyprintk and pass earlyprintk on cmdline?
<Turl>
and boot with a bootm triplet?
<oliv3r>
earlprintk cmdline!
<oliv3r>
but
<oliv3r>
without it, i should get the console prompt, no?
<oliv3r>
especially if I enable it?
<oliv3r>
earlyprintk i never had to pass before with mmc u-boot ...
boycottg00gle has quit [Remote host closed the connection]
<oliv3r>
though could be unv.bin that causes that
<oliv3r>
then again; stock kernel boots fully verbose
<arokux2>
oliv3r, are you passing console=
<arokux2>
?
<oliv3r>
arokux2: it's in my default env
<Turl>
oliv3r: if it worked, sure, but if it crashes before uart is up you won't see a thing
<oliv3r>
ok 3.4 kernel is fully verbose
<oliv3r>
i think it's crashing before uart
<arokux2>
oliv3r, well, you should to get things sorted, check wiki about booting mainline kernel.
<oliv3r>
arokux2: ...
<oliv3r>
all my mainline and 3.4 kernels boot verbose
<oliv3r>
i've booted many mainline kernels thank you
<arokux2>
oliv3r, did you do: env set fdt_high ffffffff ?
<oliv3r>
gonna try experimental/3.10 now
<arokux2>
oliv3r, don't be angry, something is wrong on your side.
<oliv3r>
it's a .config problem
<arokux2>
oliv3r, something, could be .config, could be u-boot env.
<Turl>
arokux2: backread my comments about usb
<arokux2>
Turl, did already.
<arokux2>
Turl, well, I can implement it, but EHCI isn't working 100% yet, so I'm reluctant to start new battlefield. or you want your snd card to work?
<arokux2>
granite_crusher, hi there, any news?
<oliv3r>
arokux2: well my old kernels boot verbose; but i'll boot 3.10; i know 3.10 should work, as I submitted my first patchset around that time
<Turl>
arokux2: to replace 3.4 on my mele I need it to work
<arokux2>
oliv3r, ok, no problem.
<granite_crusher>
well, that default config... I am still tweaking :)
<oliv3r>
so if that doesn't work
<oliv3r>
something is realyl fucked :)
<Turl>
arokux2: I think the port "switches" to OHCI mode, and as there's no driver it renders it unusable
<arokux2>
granite_crusher, looking for something?
<arokux2>
Turl, can you take a look which one handles it on desktop? (ohci/ehci) maybe it is ehci bug again.. and not only wifi is broken.
<granite_crusher>
arokux2: oh and I couldn't get git remote to work; so cloned to separate tree
<arokux2>
granite_crusher, how so?!
<arokux2>
granite_crusher, you are wasting too much space.
<granite_crusher>
arokux2: it reported it have unressoleved conflict...
<arokux2>
granite_crusher, hm.. you probably had some uncommited changes?
<Turl>
arokux2: [10660.657167] usb 5-1.3: new full-speed USB device number 4 using ehci-pci
<Turl>
a USB2 device does [15792.324424] usb 5-1.4: new high-speed USB device number 5 using ehci-pci
<arokux2>
Turl, can you post full config?
<hugobo>
Turl: yes, agreed!
<Turl>
arokux2: a full config for what?
<arokux2>
Turl, err. dmesg.
<Turl>
arokux2: on my PC?
<hugobo>
Turl: maybe I should reformulate my question: given an ELF executable, how do I generate a .axf?
<arokux2>
Turl, yep
<Turl>
hugobo: disassemble the boot0/1 code and figure out :)
<granite_crusher>
arokux2: don't worry about space...
<arokux2>
Turl, from the board would be nice too. btw, did you verify if storage worked? and.. my tree actually has changed, have you checked it out just now?
<arokux2>
granite_crusher, and time...
<hugobo>
I would prefer if someone knew the answer, even if I like assembly programming :)
<oliv3r>
that should be the most stable and reliable stepping stone :)
<oliv3r>
a dts kernel, that prints and has a console
<arokux2>
oliv3r, alright! let me know once you've backported everything. i'd give my lousy drivers a try :p
<oliv3r>
lol
<oliv3r>
everything will be slow
<arokux2>
oliv3r, imho, u'd better find out what the hack with (almost ready) sata driver
<oliv3r>
but 3.10 has a10 support in mainline allready
<oliv3r>
arokux2: yeah but it needs to boot first :(
deasy has joined #linux-sunxi
<hno>
oliv3r, there is a saveenv command to store changes from the command line permanently.
<oliv3r>
hno: into the nandb?
<hno>
yes.
<hno>
if that's the env partition.
<oliv3r>
hno: stock android nand so I would assume so :)
<arokux2>
hno, have some time to review my patch?
<granite_crusher>
arokux2: where to give kernel rootfs arguement (I dont want to initramfs because I need to test some peripherals as USB audio, so..) ?
<hno>
beware that saveenv was broken in some lichee releases
<oliv3r>
hno: :S ok well it's not hugely important yet; and for that 1 time change, i can always manually update env
<oliv3r>
Turl: [ 1.670000] Warning: unable to open an initial console.
<oliv3r>
hah! i overlooked that
<arokux2>
granite_crusher, you need that cpio i've uploaded, do you have it?
<granite_crusher>
arokux: why? if usb should work
<hno>
arokux2, not yet. Completely crazy at work at the moment so not much brain cycles left when at home.
<granite_crusher>
arokux2: I have it, but how I am going to test half of usb things on that tiny thing?
<arokux2>
hno, I miss you here though.
<arokux2>
granite_crusher, what do you mean? you connect them one by one...
<hno>
yes, and my email backlog is growing out of control again..
<granite_crusher>
arokux2: well ok, I will try first with initramfs
<arokux2>
granite_crusher, sorry, I didn't get your question about rootfs. you need to provide it somewhere....
<arokux2>
granite_crusher, yes, it is easiest.
<arokux2>
granite_crusher, make menuconfig
<arokux2>
Once there, fill the option General Setup / Initramfs source file(s) with a path to your initramfs image.
<arokux2>
granite_crusher, ^
<arokux2>
and do this in u-boot: env set fdt_high ffffffff
<Turl>
arokux2: maybe you notice some other mistake while implementing ohci :P
<arokux2>
Turl, or maybe I repeat same ones? :p
<arokux2>
Turl, you better push pll6, so I start pushing too. (usb clk)
<Turl>
arokux2: you can push independently
<Turl>
arokux2: maxime handles the DT and he's aware of the dependency
<arokux2>
Turl, it's clk, not DT
<arokux2>
Turl, well, both.
<arokux2>
I do not know why, but mainline feels much faster
<oliv3r>
arokux2: then what? 3.4?
<arokux2>
oliv3r, yep
<Turl>
arokux2: on what operation?
<oliv3r>
arokux2: but 3.4 is crap :
<oliv3r>
:DF
<arokux2>
Turl, logging in, issuing lsusb
<Turl>
arokux2: I still can't get lsusb to work
<oliv3r>
btw, i'm giving up for today
<Turl>
always get -99 error
<oliv3r>
Turl: i'll bug you tomorrow about [ 1.670000] Warning: unable to open an initial console.
<hno>
arokux2, life should return to normal in about two weeks. Major product launch pending at work next weekend, and then one more week to get everything shipped and the 60 racks of hosted equipment to run smootly.
<arokux2>
hno, cool.
<oliv3r>
hno: sounds awesome :)
<Turl>
oliv3r: make sure you have OF_PLATFORM_SERIAL or such
<arokux2>
Turl, even if storage works?
<Turl>
arokux2: yeah, probably missing some toggle on my kernel config
<Turl>
I should strace lsusb and see what it's missing
<arokux2>
Turl, dunno
<hno>
oliv3r, please keep track of u-boot patches meanwhile. My mailbox will overflow...
<oliv3r>
hno: okay; i'll check mailinglists and github for now
<oliv3r>
tomorrow *
<arokux2>
except of me I haven't seen any u-boot stuff
<oliv3r>
arokux2: hno ment submitted patches :)
<oliv3r>
though I don't recall seeing any recent u-boot patches
<hno>
and kernel hackers, I have a nice little task. We need to get the nand boot driver integrated/reimplemented in 3.4. See yuq work. Should be possible to add something similar to aw nand driver.
<Turl>
oliv3r: usb :p
<arokux2>
oliv3r, I got it. I consider my patch to be 80% submitted :p
<oliv3r>
arokux2: oh i should have tested that too ;)
<arokux2>
hno, who's yuq?!
<oliv3r>
arokux2: yuq write the mtd driver
<oliv3r>
hno: I think hansg/patrick wood did that allready; with the aw-nand driver; from the 3.4 SDK
<oliv3r>
hno: unless you speak of the mtd nand driver ;)
<arokux2>
hno, why should we bother about legacy 3.4?
<hno>
oliv3r, I don't think there is much pending, but worried I might miss stuff in the next two weeks.
<hno>
oliv3r, I know there have been updates to the NAND drier, but I have not seen a boot block driver added.
<Turl>
oliv3r: cini may be sending an uboot patch too for SMP on A20
<oliv3r>
hno: i'll watch the ML visously :)
<oliv3r>
Turl: he linked our u-boot github branch in a mail
<oliv3r>
Turl: i'll respond tomorrow
<oliv3r>
Turl: if you get bored; feel free to pull sunxi-3.10 and hack defconfig :)
<Turl>
oliv3r: and attached a WIP patch
<oliv3r>
Turl: ok i'll test it the next 2 weeks
<oliv3r>
SERIAL_OF_PLATFORM!
<oliv3r>
that was missing
<Turl>
:)
<Turl>
oliv3r: you got a kernel not using 8250_DW yet :p
<Turl>
probably your uart is gonna interrupt and get stuck :D good times lol
<oliv3r>
Turl: hmm?
<oliv3r>
i got 8250_DW quickrs on
<Turl>
oliv3r: yeah but your dt uses the other one :)
<oliv3r>
so 3.10.12 is probably to old?
<Turl>
oliv3r: anything older than torvalds/master is :p
<arokux2>
mnemoc, where are you help! ^^^^
<oliv3r>
Turl: LOL
<Turl>
oliv3r: need to backport all the DT fixes
<oliv3r>
Turl: yeah more work then i thought it would be to get basic support up
<oliv3r>
so OF_SERIAL is enough?
<Turl>
yeah, but it may get stuck once in a while in reboots or so
<oliv3r>
that's fine
<Turl>
(remember the uberspam serial message when rebooting some time ago on 3.4?)
<oliv3r>
it's only to test now
<mnemoc>
arokux2: oliv3r and Turl are the 3.10 fellows. I can barely focus in 3.4
<oliv3r>
ah yeah
<oliv3r>
O.o
<oliv3r>
:p
<mnemoc>
arokux2: I was only "advocating" 3.10
<arokux2>
mnemoc, oh yes, now you disappear :D
<mnemoc>
of course :p
<oliv3r>
Turl: so one more thing before going to bed; the double printing, what exactly causes that, you say earlyprintk; but i thought ywe wanted that to get early stuff printed
<Turl>
oliv3r: do you get the disabling bootconsole message now?
<mnemoc>
arokux2: next month I'll finally be in a permanent location with real internet connection and will be able to resume my sunxi work
<mnemoc>
but on 3.4
<arokux2>
mnemoc, why do you bother about 3.4?
<Turl>
mnemoc: still leeching stuff with that 3G dataplan? :(
<Turl>
arokux2: because it's usable :)
<arokux2>
Turl, but is something missing there?
<mnemoc>
because it's the stable think people can use, and what we can use to test unifications with real users
<hno>
arokux2, we bother about 3.4 because it's where we have drivers...
<Turl>
arokux2: bugs, cleaning ugly code
<arokux2>
hno, yeah, but once it works we better do not touch it.
<arokux2>
Turl, bugs +1, cleaning - no.
<arokux2>
(blush)
<mnemoc>
I want to "fake" dts spawing platform_devices from script.bin.... and so bring drivers closer to the DTS world
<Turl>
arokux2: you need cleaning
<Turl>
to notice diff between plats
<arokux2>
Turl, ah, ok
<arokux2>
Turl, good point.
<mnemoc>
arokux2: allwinner keeps a separated branch for each soc... and they make incompatible changes
<hno>
arokux2, it's software.. there will always be bugs + driver updates etc.
BJfreeman has quit [Quit: had a good time]
<mnemoc>
we first need to unify those drivers. be sure they still work everywhere. then fix bugs over a platform we know things work
<mnemoc>
to detect regressions, etc
<arokux2>
ok, mnemoc you are good advocate.
<mnemoc>
:p
* mnemoc
shuts up now
<wingrime>
mnemoc: now difficult merge 3.4 on top of 3.10?
<mnemoc>
wingrime: a lot
<arokux2>
wingrime, very difficult.
<hno>
arokux2, also, having the NAND boot block driver is a precursor to be able to implement u-boot NAND SPL loader, plus it's the last kernel piece missing to be able to make a fully open installer eleminating livesuit from the picture.
<granite_crusher>
arokux2: I got bootup and usb from which I booted uImage is visible, but I unchecked big file support (looks ext4 doesnt work without)
<mnemoc>
3.8 got multiplatform and a huge redesign of the arm world
<arokux2>
hno, +1
<mnemoc>
we need to take (unified) drivers one by one
<oliv3r>
Turl: yes, (I may have gotten it before, but overlooked it it's quite high
<steev>
well, you'll need to do arm equivalent, i'm avoiding 4.8 for the moment
<steev>
you can also look into 455bd4c430b0c0a361f38e8658a0d6cb469942b5 and 418df63adac56841ef6b0f1fcf435bc64d4ed177 - some people report those 2 fix the issue
<steev>
on arm
<arokux2>
Turl, do u wanns see something funny?
<steev>
rz2k: if you have a linus checkout, you can do something like git format-patch -o patches 455bd4c^..455bd4c
<steev>
and then 418df63
<steev>
will make a patch from each, or you can try to cherry pick
<rz2k>
455bd4c430b0c0a361f38e8658a0d6cb469942b5 is already in stage/3.4
<rz2k>
418df63adac56841ef6b0f1fcf435bc64d4ed177 too
<steev>
interesting
<Turl>
arokux2: hm?
<arokux2>
Turl, soon I'll have totally stripped down version of aw's ehci driver, which works
<steev>
rz2k: maybe another of their optimizations broke it
<steev>
again*
wingrime has quit [Ping timeout: 245 seconds]
<Turl>
arokux2: on 3.4 or 3.12?
<arokux2>
Turl, 3.4 :((
<tm512>
the sun4i_defconfig does not have DEVPTS_FS enabled?
<steev>
tm512: you mean DEVTMPFS ?
<tm512>
no, I mean for /dev/pts
<tm512>
xterm is complaining about not enough ptys
<tm512>
I might have to do legacy PTYs
<steev>
that's UNIX98_PTYS not DEVPTS_FS
<steev>
at least, afaik
<tm512>
I guess it used to be DEVPTS_FS
<tm512>
compiling a new kernel
<steev>
but, that should be in there when you do something like make sun4i_defconfig
<steev>
if you try cp sun4i_defconfig .config, it wouldn't, it would prompt you with make oldconfig
<tm512>
it has the dev/pts stuff, I am guessing that it needs legacy PTYs
<steev>
shouldn't
<tm512>
it's xterm from pkgsrc, I dunno
<steev>
unless you're on some unbelievably ancient version of xterm
<steev>
pkgsrc is... crux? arch?
<tm512>
netbsd's package system
<tm512>
that is very portable
<popolon>
is there a maintenance on wiki database ?
<steev>
oh then yes, you'll need LEGACY_PTYS since that's what BSD uses
<steev>
tm512: any particular reason you're using that over xterm that everyone else uses?
<tm512>
I guess it might assume that linux systems will work with it as well, and that since netbsd uses it, it needs it
<tm512>
and I'll be using urxvt after I get X11 up and running
<tm512>
I am doing linux from scratch
<arokux2>
Turl, what is: sw_vic?
<tm512>
hard mode, basically
<steev>
tm512: that's kind of stupid to use the netbsd version, over the actual upstream version, no?
<steev>
i wasn't aware lfs said to use netbsd's version
<tm512>
steev: no, since pkgsrc does dependency management and everything. I just did cd /usr/pkgsrc/x11/xterm && make install clean
<tm512>
and it didn't, lfs doesn't say anything about package management
rz2k has quit []
<tm512>
so I am using possibly the best choice for it. http://pkgsrc.org/
<tm512>
at least from a linux from scratch system, it's the best choice, since it handles binary and source
<steev>
*shrug* to each their own, but no, LEGACY_PTYS hasn't been enabled in a defconfig in a *very* long time
<steev>
you should probably tell the pkgsrc peeps that they should throw a warning about that, or at least a check and a warning message when you build it
<steev>
wow... pkgsrc's how to use is.... pretty much the worst how to i've ever read
<steev>
reminds me of an xda forum post
<steev>
HOWTO BUILD LINUX FOR YOUR <FANCY PHONE THAT 3 PEOPLE HAVE> 1) download this file, 2) install it via adb push 3) reboot
<popolon>
Does the wiki database administrator need some help ?
<arokux2>
popolon, why do you think so?
<tm512>
steev: that didn't fix it either, though, I'm afraid
<tm512>
"open ttydev: No such device or address"
<popolon>
arokux2, because the database is in maintenance during whole the time I translated a long article ;)
<popolon>
and my mainjob is sys & net admin
<steev>
tm512: well i've no idea. maybe check what crux does? i know they use pkgsrc or something along the lines of pkgsrc
<popolon>
sys, net, db
<arokux2>
mnemoc, ^
<tm512>
installing urxvt since it's somewhat more modern
<arokux2>
popolon, btw, I wonder why do you spent time on translations. can't french read english?
<popolon>
most french can't
<popolon>
probably an important part of computer scientists people can :)
<popolon>
and that's easier for me to improve french documentation first, and backport to english after.
<arokux2>
popolon, I see
<arokux2>
Turl, ping
<akaizen>
So only Allwinner A10 has HDMI-CEC support?
<akaizen>
(well hardware anyway)
<akaizen>
Btw who would Allwinner help more? Community with nice wiki or company who helps them sell 100,000 units?
<akaizen>
I'm trying to do the latter and hopefully make a really good product and opensource lots of it
<popolon>
Too tired, go to bed, bakcup-ed translation, will wait tomorrow to import it.
<popolon>
good night
popolon has quit [Quit: Quitte]
<akaizen>
Hmmm ... looks like A20 has a CEC pin
<akaizen>
Pin layout - P 23 HCEC
<arokux2>
I'm puzzled
<Turl>
arokux2: sw_vic is interrupt controller I think
<Turl>
arokux2: I see some delay between sie clk, phy reset and usb passby
<Turl>
your code didn't have that I think?
<arokux2>
Turl, yep... but I've tried with it...
<Turl>
arokux2: are you testing on a10 btw?
<arokux2>
Turl, yep
<arokux2>
Turl, Mele A1000
<arokux2>
these 230 LOC work.
<arokux2>
Turl, puzzling huh?
<Turl>
arokux2: did you try rtlwifi from backports on 3.4 yet?
<arokux2>
Turl, no, too much work I've though, after reading their docs
<arokux2>
thought*
<hno>
nedko, yes and yes.
<hno>
rz
<Turl>
arokux2: maybe it's dma masks or sth
<arokux2>
Turl, they are set correctly.
<arokux2>
Turl, so it is "sth", but WHAT?
<arokux2>
hno, wanna a puzzle?
<nedko>
hno: do you know if someone is already working on such driver?
<Turl>
arokux2: maybe dma itself?
<Turl>
arokux2: you can try running with full debug on both and comparing the logs, long shot though
<arokux2>
Turl, you want to say I've done the short one? :)
<arokux2>
Turl, dunno how to enable even more logs, though
<Turl>
arokux2: no, but afaik you haven't run with all those debugs on 3.4 to compare
<arokux2>
Turl, I don't understand. what do you mean?
<Turl>
arokux2: run wifi on both 3.4 with the spammy logs on and find similarities and differences
<Turl>
both 3.4 and mainline*
Amitabha has quit [*.net *.split]
wens has quit [*.net *.split]
moofree_ has quit [*.net *.split]
Tsvetan has quit [*.net *.split]
lkcl_ has quit [*.net *.split]
arete74 has quit [*.net *.split]
book` has quit [*.net *.split]
focus has quit [*.net *.split]
mdp has quit [*.net *.split]
on1x has quit [*.net *.split]
setkeh has quit [*.net *.split]
Soru has quit [*.net *.split]
ssvb has quit [*.net *.split]
techn_ has quit [*.net *.split]
buZz has quit [*.net *.split]
torindel has quit [*.net *.split]
aep has quit [*.net *.split]
zumbi has quit [*.net *.split]
tkoskine has quit [*.net *.split]
\\Mr_C\\ has quit [*.net *.split]
dlan has quit [*.net *.split]
Turl has quit [*.net *.split]
ganbold has quit [*.net *.split]
<arokux2>
Turl, as said. I cannot turn on even more logs. everything else is identical.
moofree has joined #linux-sunxi
wens has joined #linux-sunxi
Amitabha has joined #linux-sunxi
Soru has joined #linux-sunxi
ganbold has joined #linux-sunxi
arete74 has joined #linux-sunxi
on1x has joined #linux-sunxi
dlan has joined #linux-sunxi
setkeh has joined #linux-sunxi
aep has joined #linux-sunxi
book` has joined #linux-sunxi
Turl has joined #linux-sunxi
focus has joined #linux-sunxi
techn_ has joined #linux-sunxi
ssvb has joined #linux-sunxi
buZz has joined #linux-sunxi
lkcl_ has joined #linux-sunxi
mdp has joined #linux-sunxi
tkoskine has joined #linux-sunxi
Tsvetan has joined #linux-sunxi
torindel has joined #linux-sunxi
zumbi has joined #linux-sunxi
<steev>
arokux2: how is using backports too much work? you extract the tarball and run make. if cross compiling, you run make with some added environment variables