<arokux>
wens: not in our sunxi-3.4, there is one in cubieboard/cubie/sunxi-3.4, I have started to dig on it yesterday, first I want to get it working with cubieboard/cubie/sunxi-3.4
<torbenh3>
wens what do you mean with reinitilize ?
<wens>
torbenh3: not initializing gmac in u-boot
<torbenh3>
oh
<wens>
torbenh3: such as default u-boot-sunxi on cb2
eebrah_ has joined #linux-sunxi
tomee^ has joined #linux-sunxi
<wens>
I think someone screwed up the clock name for stmmac. It was named mii_clk_csr in the original patch.
<wens>
so it is actually the clock for the phy, not the gmac.
<wens>
I've been using it in the wrong way.
<torbenh3>
can you boot without network ?
<torbenh3>
i used nfs as rootfs. so i cant boot with network disabled.
<torbenh3>
if you can... activate clocktree in debugfs and look at it....
eebrah_ has quit [Ping timeout: 240 seconds]
<torbenh3>
drivers/common clock ...
eebrah_ has joined #linux-sunxi
Taduro__ has quit [Ping timeout: 252 seconds]
Taduro has joined #linux-sunxi
eebrah_ has quit [Read error: Connection reset by peer]
geecko has joined #linux-sunxi
<tomee^>
how does lcd0 relate to screen0 ?
eebrah has joined #linux-sunxi
eebrah is now known as Guest44838
Guest44838 has quit [Read error: Connection reset by peer]
focus has quit [Ping timeout: 264 seconds]
eebrah_ has joined #linux-sunxi
Taduro__ has joined #linux-sunxi
eebrah_ has quit [Read error: Connection reset by peer]
jaqmg has quit [Remote host closed the connection]
Taduro has quit [Ping timeout: 246 seconds]
Taduro has joined #linux-sunxi
Taduro__ has quit [Ping timeout: 264 seconds]
eebrah_ has joined #linux-sunxi
Taduro_ has joined #linux-sunxi
Taduro has quit [Ping timeout: 240 seconds]
focus has joined #linux-sunxi
<hramrach>
tomee^: if you select lcd for a screen in disp then lcd0 should be used
<tomee^>
hramrach: so eg for HDMI i would need fb0, disp_mode = 0 and that would use fb0 with lcd0, correct?
<tomee^>
hramrach: and if I don't intend to use the other video output (VGA or whatever else) I can leave fb1 widths/sizes unconfigured, and lcd1_para/lcd_used=0 ?
<hramrach>
isn't hdmi type 4? or does your board have a convertor? anyway, there is fex guide on the wiki
panda84kde has quit [Quit: Konversation terminated!]
<wens>
torbenh3: i traced the code. stmmac enables "stmmaceth" clock *after* setting some flags in gmac
<wens>
torbenh3: i can boot with emac in u-boot and usb ethernet for nfsroot. did that yesterday :)
<torbenh3>
hmm... enable_prepare is the first thing it does in stmmac_open.
<torbenh3>
what does it do before that ?
rz2k has quit [Ping timeout: 248 seconds]
<wens>
torbenh3: probe mdio bus...
<wens>
s/.*//
Black_Horseman has joined #linux-sunxi
<wens>
torbenh3: mdio bus is registered and probed during driver initialization
<wens>
torbenh3: can't do that without active gmac
<wens>
torbenh3: my guess is "stmmaceth" is for phy rx/tx clock
<hramrach>
and when it has been probed by u-boot already it returns reasonable results
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<wens>
hramrach: because u-boot already enabled ahb_gmac clock, so gmac is active.
<wens>
hramrach: and kernel only disables unused clocks after static drivers are initialized.
jinzo has quit [Quit: Leaving]
<wens>
allwinner driver probes mdio bus in gmac_open, after clock is enabled
<torbenh3>
wens: wrapping stmmac_mdio_register() in clk_prepare_enblae / disable ... should fix it. an that should be ok for mainline.
rz2k has joined #linux-sunxi
eebrah_ has quit [Read error: Connection reset by peer]
derethor has joined #linux-sunxi
mab_ has joined #linux-sunxi
<wens>
funny thing is DT for ST boards (spear*) don't have stmmaceth clock defined
<oliv3r>
just make sure that you involve the stmmac maintainer as it will have to work on many boards :)
<wens>
oliv3r: I prefer not to touch stmmac code, and just use the callbacks available
wolfy has left #linux-sunxi ["Ce tie nu-ti place, altuia nu-i face. Ca nu toata lumea are aceleasi gusturi. (G.B. Shaw)"]
eebrah_ has joined #linux-sunxi
naobsd has joined #linux-sunxi
mab_ has quit [Ping timeout: 272 seconds]
<wens>
oliv3r: but probably a good idea to write the maintainer anyway :)
<oliv3r>
well maybe the maintainer has some idea's too
<oliv3r>
and it really should be renamed to designware imo :)
<mnemoc>
submit the rename ;-)
eebrah_ has quit [Read error: Connection reset by peer]
<wens>
torbenh3: it's more than that. driver also probes hardware capability.
<wens>
oliv3r: maintainer works at ST :p
eebrah has joined #linux-sunxi
eebrah is now known as Guest10341
<torbenh3>
wens: enable the clk where its necessary. then send an RFC patch... to all the people involved.
<torbenh3>
its wrong to access io registers, whithout enabling the clk.
slapin has joined #linux-sunxi
<torbenh3>
and that this works on st is no excuse.
eebrah_ has joined #linux-sunxi
<wens>
i think the behavior was modified, maybe the maintainer didn't realize it.
<torbenh3>
yeah.
<wens>
looking through the history, originally it was ok if the clock wasn't defined. the driver just sent out a warning.
<wens>
after some code cleanup, it would fail.
<torbenh3>
well... its doesnt on st... so it went unnoticed. and thats why its going to be fixed, if they get aware.
<torbenh3>
maybe you could even make a case for a regression and get that patch in during the rc cycle ;D
<torbenh3>
probably not.
Guest10341 has quit [Ping timeout: 272 seconds]
<wens>
i'll mail the maintainer first
<oliv3r>
wens: yeah, so it's a hard sell, but in all fairness, it IS designware IP :)
<oliv3r>
torbenh3: You seem to be quite knowledgeable yet quite new around here, may I be so bold and ask who you are? :) (In the most positive sense of course)
<torbenh3>
oliv3r: well... the files are already called dwmac1000*
arete74 has quit [Read error: Connection reset by peer]
<torbenh3>
oliv3r: i work at linutronix.
<torbenh3>
oliv3r: my boss is thomas gleixner.
arete74 has joined #linux-sunxi
<torbenh3>
hmm... other than that... i spent a lot time with linux audio stuff. but since 2.5 years i am mostly doing kernel work.
<torbenh3>
and i just bought a CT 2 weeks ago. and so i just want to contribute a bit in my free time.
<torbenh3>
but that time is pretty limited. iE didnt make it to test your u-boot stuff yesterday.
<torbenh3>
hmm... i am 37 and live in bremen, germany. and i am windows free since about 1998...
ganbold_ has joined #linux-sunxi
pacopad has joined #linux-sunxi
<wens>
someone actually refers to it as dwmac on netdev ML
Black_Horseman has quit [Ping timeout: 246 seconds]
Taduro_ has quit [Read error: Connection reset by peer]
<arokux>
who knows where is wip of levitan... the guy who worked on mmc for mainline?
eebrah_ has joined #linux-sunxi
Taduro has joined #linux-sunxi
eebrah_ has quit [Read error: Connection reset by peer]
mab_ has joined #linux-sunxi
<arokux>
tomee^: hi
<arokux>
tomee^: are you running cubietech's kernel? cubieboard/cubie/sunxi-3.4?
eebrah_ has joined #linux-sunxi
Taduro has quit [Ping timeout: 246 seconds]
mab_ has quit [Remote host closed the connection]
Fusing has joined #linux-sunxi
eebrah_ has quit [Ping timeout: 252 seconds]
eebrah_ has joined #linux-sunxi
EGM has joined #linux-sunxi
<EGM>
Hi everyone.
<EGM>
Could anybody hint me what we need to do in order to get A20 working with 2GB of RAM? We have custom board running Android, and we replaced 1GB ram with 2GB.
<EGM>
We suppose that we need to replace bootloader or change it's settings.
<EGM>
It's currently using Allwinner boot1 and Allwinner u-boot. It seems like boot1 has hardcoded support for 1GB only.
<mnemoc>
livesuit should "fix" boot0 for you
<mnemoc>
boot1*
<EGM>
Sorry, boot0
<EGM>
Instead of boot1, there is u-boot.
<mnemoc>
if using our u-boot from mmc, look at the cubietruck board as reference
<EGM>
I hope I understand it correctly.
vicenteH has quit [Ping timeout: 252 seconds]
<EGM>
I believe that replacing both bootloaders with uboot-spl should solve the issue, but I am not sure. Is it possible?
<libv>
EGM: so you guys built your own board, and are now asking us for support, surely allwinner can help you out ;p
<EGM>
libv: I am not sure if you are trying to be sarcastic or just mean, but I don't see anything wrong with asking you for help.
eebrah_ has quit [Read error: Connection reset by peer]
<libv>
mnemoc: there will be a time where linux-sunxi.org should demand a royalty from allwinner on every chip sold ;)
<EGM>
Guys here helped me before, so I just asked here again.
<libv>
EGM: most of that was just an observation, the allwinner part was sarcasm
<EGM>
torbenh3: Thanks for the hint, I will look into it.
<EGM>
libv: Oh, ok.
eebrah has joined #linux-sunxi
eebrah is now known as Guest18385
<mnemoc>
libv: :)
<mnemoc>
EGM: the sys_config.fex file has a [dram_para] section, livesuit fills some of the blanks when flashing the board using some autodetection voodoo
<EGM>
mnemoc: Oh, interesting. But I am not quite sure if boot0 respects values in FEX file. It looks like there is hardcoded 1024MB.
<tomee^>
arokux: no
<torbenh3>
mnemoc: maybe you should clarify, whether the uboot detected memory is actually relevant for the fex stuff. i only know mainline ;D
<tomee^>
arokux: I mean yes I have it installed as well. But I am running linux-sunxi/stage-3.4
<mnemoc>
EGM: not those in the script.bin file, but livesuit has it's own methods for that
<mnemoc>
torbenh3: uboot passes the hardcoded meminfo to linux
<mnemoc>
torbenh3: u-boot doesn't detect anything
<arokux>
tomee^: oh, ok :) so you've compiled this --> cubie/sunxi-3.4?
<EGM>
mnemoc: So you are telling that LiveSuit should automatically adapt boot0 for memory size?
<mnemoc>
torbenh3: the lichee u-boot queries the dram controller to know what's the ram size
<mnemoc>
EGM: yes
<torbenh3>
mnemoc: hmm... oliv3r u-boot does use get_dram_size()....
<tomee^>
arokux: no, used prebuilt from "linaro xbmc" I guess. which is a copy from cubietech's "lubuntu".
<EGM>
mnemoc: Oh, ok. I am affraid that we are unable to use livesuit with that board. It does not detect the board.
<mnemoc>
EGM: look at the livesuit version used by the cubietruck
<arokux>
tomee^: ok. I'm just wondering if I could reproduce this build, will take a look at it tonight. I've started to work on adding wifi to ours sunxi-3.4
<mnemoc>
EGM: you can also use a binary editor and modify boot0
<EGM>
mnemoc: Oh, good idea. Thanks.
<arokux>
EGM: do you still have usb problems?
<tomee^>
arokux: you talking about the egl bug?
<arokux>
tomee^: egl? no
<tomee^>
arokux: oh, ok, nevermind ;)
<tomee^>
arokux: btw. does the bootloader in nand influence stuff when booting from usd?
<EGM>
arokux: I haven't tested it with recent commits, so I am not sure.
<arokux>
tomee^: no
<tomee^>
arokux: ok, so the SoC just tries uSD if its inserted, doesn't go through nand.
<EGM>
mnemoc: Is it possible to use phoenixsuit instead? I am not sure if livesuit and phoenixsuit is same software just for different OS, or not.
<tomee^>
EGM: it *seems* to be the same software, although some features are only available in newer phoenixsuit for windows.
<arokux>
tomee^: give our wiki some love, it deserves it!
<EGM>
mnemoc: I am looking at cubieboard web and there is PhoenixSuit 1.0.6, seems like we use 1.0.7.
<tomee^>
arokux: I've skimmed through it dozens of times and read tens of articles. Look at the surge of IPs from Poland in the last 2 weeks ;)
<tomee^>
arokux: didn't read THAT ONE article though.
<arokux>
tomee^: by giving some love i"ve meant "improve" it :)
<tomee^>
arokux: ha, not yet feeling competent enough, but close.
<mnemoc>
EGM: first get things working using purely allwinner's SDK
<mnemoc>
EGM: and contact them about configuration troubles using their autodetection voodoo
<EGM>
mnemoc: We will contact them, thanks a lot for information so far!
Fusing has quit [Ping timeout: 272 seconds]
<oliv3r>
EGM: tell them you want their configuration VOOODOO and then put it on our wiki ;)
<EGM>
oliv3r: I will try to achieve that :)
<EGM>
How much different are u-boots for different boards? Is it just matter of dram settings? In other words, would be possible to boot u-boot for Cubietruck on our board (A20, 2GB)?
<oliv3r>
torbenh3: get_dram_size 'tests' if your ram is valid, so if you say hey, I have 10 GiB of ram, but only 1 G installed; it should puke and error out after 1 G
codingrobot has quit [Ping timeout: 272 seconds]
codingrobot has joined #linux-sunxi
<torbenh3>
oliv3r: ok. seeing now, that the SDRAM_SIZE define is not usedi in board/sunxi/dram_cubietruck.c
shineworld has joined #linux-sunxi
rz2k has quit [Ping timeout: 246 seconds]
<oliv3r>
torbenh3: i don't think it's anywhere used, it jus tpasses .size (from the top of my head)
<oliv3r>
EGM: you can boot any sun4i u-boot on a sun4i board, there's no differences there, memory timings are board specific, but if you take a safe low clk, your usually 'safe' iowidth relates to the number of chips, so that's something to potentially keep in mind
<oliv3r>
EGM: e.g. cubietruck is sun7i, your A20 board will run fine with the same u-boot (except for the printed string of course)
<oliv3r>
EGM: that is assuming you also use 4 memory chips and can run at 436 MHz i think itss at?
<EGM>
oliv3r: Yeah, I am already trying that and it seems to work partially. Unfortunately it still tells "DRAM: 1024MB"
<EGM>
We use only 2 memory chips.
<oliv3r>
then you need a new u-boot
<oliv3r>
can you compile your own u-boot?
<EGM>
But I am but confused about dram configuration for cubietruck. While the cubieboard2 uses only 2 memory chips, and cubietruck uses 4 memory chips, there is only difference in clk, density and size.
<EGM>
oliv3r: I just compiled latest commit from u-boot-sunxi
<oliv3r>
EGM: you got some high density chips then :)
<oliv3r>
EGM: btw, memory io-width could actually be wrong for a lot of boards
<oliv3r>
i think the cubietruck actually needs a io-size of 8-bit, but hno knows this much better, I always forget how it works out exactly
<binaryferret>
The one that I attempt to load now says 'driver [ekt2132] using legacy suspend method' and same message but with resume method.
<EGM>
There is density of 8192 for CT. I assume that's in Mb, so that's 1GB per chip. But there are 4 chips.
<oliv3r>
i thought it was 1 chip, io-size is 32bit or both bus and io are 16 bit
<oliv3r>
2 chips would be io-size of 16 and a bus-width of 32
<oliv3r>
and 4 chips would be 8 io, 32 bus
<oliv3r>
EGM: we extract those settings from the dram controller with boot0/boot1 booted board
<oliv3r>
and the dram controller is completly undocumented, we just try to do what we think it means
<EGM>
oliv3r: Thanks for the info, I will mess around with u-boot for a while and try different settings.
<oliv3r>
EGM: feel free to send your final findings to the ML as a patch for inclusion
<oliv3r>
but my first guess, would be double the density
<oliv3r>
and half the io-width, though the io-width might not do anything directly
<EGM>
oliv3r: Yeah, it makes sense. But diff between dram settings for CB2 and CT does not make sense.
<oliv3r>
EGM: misterious dram controller is misterious
<EGM>
I believe it is mysterious :)
<oliv3r>
it is indeed ;)
<oliv3r>
hence, we'd love to get our hands on that voodoo livesuit does
<mnemoc>
well... there are dumps
<oliv3r>
livesuit dumps? of the sourcE?
<mnemoc>
protocol
<mnemoc>
with the stuff they upload to "probe" the dram
<mnemoc>
via fel
<oliv3r>
ah, ok, true
<mnemoc>
but even if it makes sense to probe the dram, we can't do it on every boot
Guest18385 has quit [Read error: Connection reset by peer]
<oliv3r>
EGM: btw, what libv ment, and it's fair, we aren't really Allwinners support channel; we help where we can, but technically, this question was a pure an allwinner support question ;)
<EGM>
oliv3r: I understand that. And I will try to give something back and provide at least feedback and bug reports, it not patches.
<oliv3r>
EGM: btw, our u-boot can't boot from nand
rz2k has joined #linux-sunxi
<EGM>
oliv3r: Oh, really? Why?
eebrah has joined #linux-sunxi
<oliv3r>
we have no nand controller ;)
eebrah is now known as Guest93247
<oliv3r>
there's some highly experimental MTD work being done, but it needs some love
<oliv3r>
EGM: in any case, i'd love to be able to say 'fill this into this field and that into that field and it works' but we kinda just do trial and error (or probe a pre-configured DRAM controller, which may actually be wrongly configured, but appears to work)
<EGM>
oliv3r: Oh, ok. So I took dram settings for CT, doubled the density and I got "DRAM: 0 MiB" :)
<oliv3r>
but does it boot ;)
<oliv3r>
probably not, try io-width of 8 then
<EGM>
Yeah, it does something.
<oliv3r>
it could be a printing bug
<oliv3r>
SPL and u-boot both print the dram, both say 0?
<EGM>
Actually only SPL printed something
<oliv3r>
yeah, then it fails
<oliv3r>
SPL inits dram, then loads u-boot into DRAM
<oliv3r>
since it can't, it won't work
codingrobot has quit [Ping timeout: 264 seconds]
<oliv3r>
hmm, either mmc fails to read, or it tries to write into dram and that fails
codingrobot has joined #linux-sunxi
<oliv3r>
EGM could also be, that the dram controller doesn' tsupport your chip density ;)
pfdm has joined #linux-sunxi
rz2k has quit []
vicenteH has joined #linux-sunxi
<libv>
heh... someone complained that we are not supporting all features of random mali libs out there
<libv>
in sunxi-mali
<libv>
our headers are not supporting dma_buf yet!
<libv>
oh no!
codingrobot has quit [Ping timeout: 264 seconds]
<libv>
i only plucked the headers off of khronos and then manually added the necessary bits to make mali go, and now we have become the defacto canonical source for all mali binary headers.
<EGM>
oliv3r: I messed around with density, io, bus width, etc. But still nothing with 2GB reported.
<EGM>
And considering the fact that cubietruck uses 4x512 chips, it's bad.
dwilkins has quit [Read error: Connection reset by peer]
dwilkins has joined #linux-sunxi
<swabbles>
libv: instead of appreciating all the work that has been done thus far :).
shineworld has quit [Quit: Leaving]
Fusing has joined #linux-sunxi
<Sonic1_>
hehe
Sonic1_ is now known as Sonic1
Guest93247 has quit [Read error: Connection reset by peer]
eebrah has joined #linux-sunxi
eebrah is now known as Guest64608
FDCX has joined #linux-sunxi
Guest64608 has quit [Ping timeout: 252 seconds]
ZetaNeta has joined #linux-sunxi
eebrah_ has joined #linux-sunxi
Taduro_ has quit [Read error: Connection reset by peer]
n01 has quit [Ping timeout: 246 seconds]
Taduro_ has joined #linux-sunxi
rz2k has joined #linux-sunxi
<libv>
heh, nm, this is really someone without much in the way of a clue
deasy has joined #linux-sunxi
deasy has quit [Remote host closed the connection]
eebrah__ has joined #linux-sunxi
Taduro_ has quit [Read error: Connection reset by peer]
Taduro has joined #linux-sunxi
rz2k has quit []
eebrah_ has quit [Ping timeout: 265 seconds]
techn_ has joined #linux-sunxi
pacopad has quit [Quit: pacopad]
techn__ has quit [Ping timeout: 265 seconds]
tomee^ has quit [Quit: I Quit.]
Taduro has quit [Read error: Connection reset by peer]
Taduro_ has joined #linux-sunxi
Taduro__ has joined #linux-sunxi
foubarre has joined #linux-sunxi
<foubarre>
Hi. Can anyone tell me if module g_mass_storage is active or can be activated on some cubieboard A20 distributions?
Taduro_ has quit [Ping timeout: 264 seconds]
n01 has joined #linux-sunxi
<foubarre>
or at least can you people please do a "modprobe -l"? If g_mass_storage is listed i'd like to know what is your linux.
rz2k has joined #linux-sunxi
popolon has quit [Quit: Quitte]
_whitelogger has joined #linux-sunxi
foubarre has quit []
Taduro has joined #linux-sunxi
Taduro__ has quit [Read error: Connection reset by peer]
atiti has quit [Ping timeout: 245 seconds]
deasy has joined #linux-sunxi
Taduro has quit [Quit: No Ping reply in 180 seconds.]
Taduro has joined #linux-sunxi
Sonic1_ has joined #linux-sunxi
Sonic1 has quit [Remote host closed the connection]
<tomboy64>
well, at least he held out 12 minutes
deasy has quit [Remote host closed the connection]
ZetaNeta has quit [Ping timeout: 272 seconds]
Taduro___ has joined #linux-sunxi
Taduro has quit [Ping timeout: 246 seconds]
Seppoz has quit [Ping timeout: 264 seconds]
AreaScout has joined #linux-sunxi
Taduro_ has joined #linux-sunxi
Taduro___ has quit [Read error: Operation timed out]
iamfrankenstein has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
Taduro_ has quit [Read error: Connection reset by peer]
Taduro has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
Taduro___ has joined #linux-sunxi
eebrah__ has quit [Ping timeout: 240 seconds]
dack has joined #linux-sunxi
Taduro has quit [Ping timeout: 272 seconds]
eebrah has joined #linux-sunxi
eebrah is now known as Guest38874
Taduro_ has joined #linux-sunxi
Taduro___ has quit [Ping timeout: 264 seconds]
Taduro_ has quit [Read error: Connection reset by peer]
Taduro has joined #linux-sunxi
wolfy has joined #linux-sunxi
wolfy has left #linux-sunxi [#linux-sunxi]
EGM has quit [Remote host closed the connection]
FR^2 has quit [Quit: Connection reset by peer]
pfdm has quit [Ping timeout: 250 seconds]
Sonic1 has joined #linux-sunxi
Sonic1_ has quit [Ping timeout: 264 seconds]
codingrobot has quit [Ping timeout: 264 seconds]
codingrobot has joined #linux-sunxi
Guest38874 has quit [Ping timeout: 245 seconds]
<kriegerod>
i found that linux-sunxi states no cubietruck onboard wifi support yet, while github.com/cubieboard2/linux-sunxi repo has it in sunxi-3.4-ct-v101 branch. I will test it tomorrow. I would be happy to help merge it into linux-sunxi, i'll try to prepare pull request. I am also able to test this issue (and other cubietruck-related issues).
dack has quit [Quit: No Ping reply in 180 seconds.]
<arokux2>
wens: I've tested your tree. check the dmesg at my page.
<arokux2>
kriegerod: yes, you've understood everything correctly, I'm also working on this.
<arokux2>
kriegerod: however you should use ithub.com/cubieboard/linux-sunxi cubie/sunxi-3.4 as reference, this is the up-to-date kernel that is used by Cubietech