<mouchon2>
just a question i reading spi code and try to enable it for sun7i and if i understand there is currently no dma support for sun7i ?
<oliv3r>
mripard: well it IS the cubieboard3, marketing name: cubietruck; but I did have it down as ct so far, so i'll commit it
<mouchon2>
does my understanding is correct or i missed something ?
<oliv3r>
mouchon2: there is afaik
<oliv3r>
but not all drivers are there for sun7i yet
popolon has quit [Quit: Quitte]
<mouchon2>
ok but looking at dma_defs.h used by spi_sunxi.c there is no def for sun7i should i need to add it before or need to look somewhere else ?
<oliv3r>
i think the dma controller is identical
<oliv3r>
you can look for hansg's patches in the history to see what he did to port several drivers over
cajg has quit [Ping timeout: 252 seconds]
<mouchon2>
ok thanks oliver
atiti has quit [Ping timeout: 264 seconds]
cajg has joined #linux-sunxi
ykchavan has joined #linux-sunxi
tinti has quit [Ping timeout: 240 seconds]
<Turl>
mripard: ping :)
FergusL has left #linux-sunxi ["Quit"]
<Turl>
mripard: is this reset controller suitable for everything?
<mripard>
everything like what ?
<Turl>
mripard: arokux2 has some reset "clocks" for his USB driver
<Turl>
it's just a bit on the clock register
<Turl>
maybe this would be a better way to do the reset?
<mripard>
yep
<Turl>
arokux2: ^ :)
hramrach has quit [Ping timeout: 240 seconds]
<Turl>
mripard: git am is complaining your first patch has trailing whitespace
<mripard>
fuuu
hramrach has joined #linux-sunxi
<mripard>
fixed
tzafrir_laptop has quit [Ping timeout: 246 seconds]
tinti has joined #linux-sunxi
<hramrach>
hmm, I guess there is a problem with the sdboot thingy
<hramrach>
plain card works ok
<Turl>
mripard: patches pushed to sunxi-devel :)
tinti has quit [Remote host closed the connection]
<hramrach>
do I only need bin/fel-sdboot.sunxi or more stuff?
<hramrach>
the card just does not boot with that thing written to it
<oliv3r>
erm it should force the CPU into fel-modde
<hramrach>
it boots from nand
<oliv3r>
written to the wrong spot of the sd card?
<hramrach>
it says to write at 8k
<oliv3r>
i never used it; i just force it into fel mode from a uart by pressing '2'
<oliv3r>
yeah seek=8; bs=1024
<hramrach>
I don't want to press 2 every time I boot
<Turl>
hramrach: if you can, I'd use tftp with an uboot only card
<Turl>
faster than fel fwiw
<hramrach>
technically I can I guess
<hramrach>
if u-boot supports gmac or wht it's called
<Turl>
oliv3r: get to work! ^ :p
_whitelogger has joined #linux-sunxi
<wingrime>
oliv3r: what is TSD
rz2k has joined #linux-sunxi
<Turl>
wingrime: it is "SD" with NAND inside
<Turl>
bbl
<wingrime>
we can 64gb sd cards?
<djahma>
Compiling various sunxi kernels, the best result I can obtain is for my tablet is touchscreen working. Wifi and BT still unavailable. So here is the list of modules as used by android: http://pastebin.com/Xm5FKRH1 Could someone help to get this wireless part working on my A10?
mouchon2 has quit [Ping timeout: 245 seconds]
<oliv3r>
hramrach: gmac + phy doesn't work yet in u-boot; it's falling back to the emac with the gbit PHY an though in theory that should work; i need to see why it won't. or rather, i have no clue so I should ask someone that might know
<oliv3r>
wingrime: turl lies, it's an embedded MMC chip. e.g it looks like a nand chip, but connects like a SD card. the samsung galaxy s2 and s3 have it as their main storage. it's like a permanently mounted sd card
<wingrime>
oliv3r:
<wingrime>
eMMC
<wingrime>
usualy named
<arokux2>
what is a reset controller? o_O
<arokux2>
mripard: why do you select ARCH_HAS_RESET_CONTROLLER for all sunxi, if only sun6i has it?
<oliv3r>
embedded MMC yes, but the chips are named TSD for some reason; don't ask me why
<oliv3r>
arokux2: erm, all sunxi's could have a reset function, the sun6i has dedicated registers for it, for other arches, we may have to hack in stuff for it
<oliv3r>
but the reset framework is good (tm)
<arokux2>
oliv3r: could have? so you do not know. why are you selecting that they all have it then?
mouchon2 has joined #linux-sunxi
<arokux2>
I suppose because there is no ARCH_SUN6I
rz2k has quit []
ykchavan has joined #linux-sunxi
<hramrach>
djahma: btusb is generic bt over usb. did you compile that into the kernel? I don't think it's in anyone's config. you have to enable it
mouchon2 has quit [Read error: Operation timed out]
<hramrach>
same for the wifi whatever it is. the defconfigs only enable wifi known to ship with some device. you have some different wifi so you will have to enable it in config.
<hramrach>
djahma: maybe if you pasted output of dmesg commant it would be obvious what wifi you have
eebrah|noPC has joined #linux-sunxi
<hramrach>
oliv3r: probably eMMC is a trademark and somebody did not want to pay royalties so TSD :p
<djahma>
hramrach: let me try dmesg over adb shell and see what wifi this tablet is using
<hramrach>
djahma: yes, rtl8150 is normal usb wifi. just enable the driver and rebuild the kernel
<hramrach>
I guess it's not in defconfig because it's not required minimum to boot a board
gzamboni has joined #linux-sunxi
<hramrach>
and if you have the module and it does not work then .. well, you could complain to upstream maintainer if you were using non-obsolete kernel
<hramrach>
actually it might be rfkill on a gpio which turns on power to the wifi and when you don't have driver for that it won't work
popolon has joined #linux-sunxi
<hramrach>
djahma: do you have that module and no actual wifi card on usb or are you missing hte module?
<djahma>
hramrach: what I pasted on pastbin is a list of drivers from android, using adb shell. Now I am trying to figure out what wifi card is used by the tablet to get it working under linux. To answer your question, there's nothing plugged in any usb port; only a mouse or keyboard from time to time when under linux
<djahma>
the thing is I can't log in android: "insufficient permissions for device" to run dmesg. This is odd, everytime before I just needed to do "sudo adb shell" to get in...
<djahma>
ok so I cant run that command, don't know why I cant get into android anymore
mouchon2 has joined #linux-sunxi
atiti has joined #linux-sunxi
<arokux2>
Turl: your sunxi-clk branch has the patches you have sent as they are?
<djahma>
so is the correct driver for both BT and wifi : BCM40183?
akaizen has quit [Remote host closed the connection]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux2>
Turl: you there?
<Turl>
im afk
<Turl>
dunno if they are the exact same, diff them :)
<Turl>
or use sunxi-devel
eebrah|noPC has quit [Ping timeout: 240 seconds]
rellla has joined #linux-sunxi
<djahma>
hramrach: YES! I think I understand what's going on with this wifi chip! :-) basically, the original android kernel was compiled with bcm40183 module which uses both a firmware (bcm40183_fw.bin) and a nvram (bcm40183_nvram.txt) whatever that is. So I must compile the sunxi kernel with the BCMHD module and pass it the path to both these files as argument. However, I am new to cross compilation and I would need some guidance there
iamfrankenstein has quit [Ping timeout: 240 seconds]
<djahma>
Now, when I look into the tablet files, I see a folder called vendor_modules with variations on this bcm40183: I have files for bcm40181 and bcm4330. modinfo on them indicate they all connect through GPIO so maybe be I could use any of them? (wifi signal on android was very weak, maybe a better module would get me a better signal?)
<djahma>
also, the sunxi kernel only offers BCMDHD which is for 4329/30 wireless cards, so I could not compile it for bcm40183
<djahma>
what do you think is the proper way to do it?
<djahma>
and how to give the module the proper paths? are the path the places where the kernel once compiled will search, in the tablet? or are these path the places where I currently have these files in my computer?
<djahma>
(I've been trying to get that working for 2 months now! That's getting exciting ;-) and yes I'm a newbie in that field
rellla has quit [Remote host closed the connection]
zetaneta has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
<arokux2>
Turl: still afk?
akaizen has quit [Ping timeout: 245 seconds]
atsampson has quit [Ping timeout: 240 seconds]
<Turl>
arokux2 I'll be afk until tue
<arokux2>
Turl: ok
<Turl>
typing on phone is slow :)
deasy has joined #linux-sunxi
atsampson has joined #linux-sunxi
<arokux2>
Turl: I think I've found a bug in sunxi_gates_clk_setup
<arokux2>
Turl: hm.. but kernel won't boot then.. strange.
<Turl>
no, i is right
<Turl>
because you index by hw offset (i), not sequencial (j)
<arokux2>
Turl: but clks is just an array?
<Turl>
so?
<arokux2>
Turl: a sequential array of children clocks
<Turl>
it neednt be continuous
<arokux2>
Turl: so index should be sequential. why there should be holes in that array?
<Turl>
it is not seq
<arokux2>
Turl: by design?!
<arokux2>
Turl: by the requirement of the clk framework I mean?
<Turl>
holes need to be because holes exist in the register
<arokux2>
Turl: isn't it enough just to pass a bit number to the clk_register_gate ?
<Turl>
for it to work, yes
<arokux2>
Turl: so who exploits the fact there are holes in clk_data->clks then?
<Turl>
buy you lose the 1:1 <&gates bit> with register bit if continuous
<Turl>
mapping
<Turl>
dt
<arokux2>
Turl: now it makes sense. but I do not like it. is it a standard practice to do so?
<Turl>
yes
<steev>
ssvb: hm, just noticed the github message - i'm not sure what you mean exactly? i haven't added the sunxi stuff yet, as you were covering it. i've been thinking about it though
<arokux2>
Turl: ok.. this is odd. to know what ind to put at <&usb ind> I need to check the mask...
<steev>
i would love to add stuff to the tree, but right now with most of them requiring "hacks" to mesa and whatnot, kind of hard to
<Turl>
usb does not offer a bunch of gates does it?
<Turl>
btw you should use reset framework for those reset bits i think
<arokux2>
Turl: it does, so I'm using this general func now and it failed to work... because of this oddness...
<arokux2>
Turl: hm.. I know nothing about what this reset framework is good for.
<Turl>
read the backlog with mripard
<Turl>
and ask him :)
<arokux2>
Turl: do you know that usb can have something in common with this framework? maybe it was just unfortunate name for the clocks?
<Turl>
isnt it 1 bit you trigger to reset the ip?
zetaneta has joined #linux-sunxi
zetaneta has quit [Changing host]
zetaneta has joined #linux-sunxi
<arokux2>
Turl: the chinese code handles it as clock.
<arokux2>
Turl: and this is in __CCMU_USB_CLK_REG00CC module.
<arokux2>
Turl: do you still think it has something to do with reset?
<arokux2>
mripard: you there? I need some help :)
jinzo has joined #linux-sunxi
<oliv3r>
Turl: cubietruck dts is sent
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux2>
oliv3r: you seem to know about reset framework?
djahma has quit [Ping timeout: 264 seconds]
ykchavan has quit [Quit: Leaving]
djahma has joined #linux-sunxi
djahma has quit [Client Quit]
<oliv3r>
arokux2: the reset framework is just a framework you register against, to easily and quickly allow you to reset your device, as far as I understood it anyway
vicenteH has joined #linux-sunxi
<oliv3r>
the a31 does actually have reset registers to reset components on the hardware level
<arokux2>
oliv3r: ok, any idea where mripard got info on these registers ?
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<oliv3r>
arokux2: aw secrets :)
<arokux2>
oliv3r: wtf?!
wingrime has quit [Ping timeout: 240 seconds]
q3k has joined #linux-sunxi
<q3k>
huh. I'm trying to get my A20 to boot from NAND, but i get this error: "boot1: bad boot image magic, maybe not a boot.img". I barely touched the /dev/nanda partition, I een restored it from a stock image.
<q3k>
Yet I still get this error.
<q3k>
I am just about to reflash the board with livesuit, but I already installed a distro on NAND...
iamfrankenstein has joined #linux-sunxi
granite_crusher has quit [Quit: Page closed]
<oliv3r>
you can always reflash with lifesuit
<oliv3r>
but boot1 fails to load your u-boot.bin
<q3k>
yes, that's what I'm about to do, I'll backup my current NAND first.
<q3k>
and yes, I'm also confused by the situation. if I change the address in linux.ini, it doens't get reflected in the boot messages....
<q3k>
so something is seriously messed up. yet, when I format the /dev/nanda partition, I get a boot.axf missing error. so it does get read.
<oliv3r>
linux.ini i think does get read by boot1 but not sure
<q3k>
anyway, I'll just reflash it. :(
<oliv3r>
well nanda is just a fat partition
<oliv3r>
you can in theory just copy back its files
<q3k>
well, that's what i tried. I extracted a stock android image with awimage, copied the files over (then even dd'ed the partition), but I still get the same error.
<q3k>
so I dunno :<
<oliv3r>
'stock android image'?
<oliv3r>
or one specific to your tablet
<oliv3r>
dd probably fails for teh same reason copy failed
<oliv3r>
the file signature is wrong :)
<q3k>
stock android image for my cubieboard2 :>
<oliv3r>
btw, boot1 loads boot.axf i think
<oliv3r>
so that could be corrupted?
<q3k>
I don't think I ever touched it...
<oliv3r>
well in case of trouble, just restore the entire thing :)
Seppoz has quit [Remote host closed the connection]
Seppoz has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 240 seconds]
Gerwin_J has quit [Quit: Gerwin_J]
Seppoz has quit [Ping timeout: 260 seconds]
deasy has joined #linux-sunxi
<arokux2>
hno: what does it mean to "leave a chip in reset"?
Seppoz has joined #linux-sunxi
<oliv3r>
my guess, is to have it in a 'off' position, but ready to go on fast
<oliv3r>
like when you keep the reset button of your computer pressed :) when you release, it instantly is on
<arokux2>
oliv3r: this is something basic for arm architecture
<oliv3r>
aye, i tried to make a simple comparission
<oliv3r>
but if you look at microcontrollers, they all have a 'reset line'
<oliv3r>
the time from when you assert that line, to the actual reset, takes several ms
<arokux2>
oliv3r: asserting means write 1 to it?
<oliv3r>
so my guess is, that it does the actual reset, but waits for it to 'run' again
<oliv3r>
'hold low' or 'hold high' i think
<oliv3r>
but that's just a guess :)
<oliv3r>
a welle ducated guess *
<arokux2>
I cannot find any info on it. where should it be?
<oliv3r>
probably doesn't exit
<oliv3r>
(for aw socs)
<arokux2>
oliv3r: not for aw, in general
<oliv3r>
we still strongly belive, that the manuals get written after the soc :)