<ganbold>
arokux2: rootfs on usb stick works on cubieboard1,2 (FreeBSD)
<arokux2>
ganbold, oh.. hi
<arokux2>
ganbold, have you seen any quirks? :(
<arokux2>
ganbold, I just cannot get it working...
<ganbold>
arokux2: I didn't try linux so I can't tell
<arokux2>
ganbold, so you have written it down and it worked? :)
<ganbold>
arokux2: I wrote glue code for a10 ehci and it worked
<ganbold>
it works for a20 too
<arokux2>
ganbold, yes, i'm working on this glue too...
<ganbold>
arokux2: if host mode works and you can mount usb stick after booting linux, then maybe rootfs related problem
<arokux2>
ganbold, btw, do you have an understanding what the glue code does or you just copied it?
<ganbold>
arokux2: I looked through linux-sunxi codes, wait a moment
<ganbold>
arokux2: first in a10_ehci.c look at _attach function, there should be lines that first activates clock then gives power to usb, then later enables passby, configure port codes
<ganbold>
arokux2: what doesn't work for your case?
<arokux2>
ganbold, well, it is impossible to say :(
<arokux2>
ganbold, I've found a few bugs in my porting, but there might be some more
<ganbold>
arokux2: can you mount usb stick?
<arokux2>
ganbold, no, on my board there is an additional hub, it gets identified, but I get errors
<ganbold>
arokux2: what board you have?
<arokux2>
ganbold, Mele A1000
<ganbold>
arokux2: somebody in freebsd-arm mailing list was able to run freebsd on mele
<arokux2>
ganbold, nice
<arokux2>
ganbold, you see, most probably my code is buggy, but as I do not understand what all those things are I do not know what fails
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux2>
ganbold, for some reason I get csr3: usbc1: 0x72a80, but sunxi code gets 0x22b80.
<ganbold>
arokux2: maybe you can try commenting lines and enabling them line by line, linux-sunxi code is mess (only in my opinion), so if it is easier for you then you can look freebsd a10_clk.c and a10_ehci.c codes, I think it is much less
<arokux2>
ganbold, yes, you are right, I actually cleaned it and went through all this mess
<arokux2>
ganbold, anyway, nobody can help me by myself.. :)
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<ganbold>
arokux2: don't worry much, I started from zero porting freebsd to cubieboard and it took me 3 months
<arokux2>
ganbold, :)
<arokux2>
ganbold, thanks for support
<ganbold>
arokux2: guess how many times I compiled kernel :)
<arokux2>
ganbold, yeah, same here. I've set up NFS so it is much faster, do not need to fiddle with sd cards
<ganbold>
arokux2: good, I don't have working ethernet driver yet :) of course I could try usb ethernet though
<arokux2>
ganbold, get ethernet working first this will pay off
<arokux2>
ganbold, ethernet was there in my case..
<tm512>
I guess I need to prepare an SD card and specify a resolution for this to work?
\\Mr_C\\ has quit []
<Turl>
vinifr: what about it? :)
\\Mr_C\\ has joined #linux-sunxi
<vinifr>
Turl, Are you making an improvement? Or adding something?
<Turl>
vinifr: I'm the driver author
<vinifr>
Turl == Emilio Lopes??? :)
<Turl>
vinifr: yep
Seppoz has quit [Remote host closed the connection]
akaizen_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
vinifr has quit [Quit: Saindo]
<wens>
arokux2: mripard's new branch works
<arokux2>
wens, alright!
<wens>
arokux2: my config still needs work. I can't get init to run. something is missing.
<arokux2>
ok, have you included ELF support? (heard even this one is off for allno)
<wens>
arokux2: yes there is :p
Amitabha is now known as amitabha
amitabha is now known as Amitabha
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
<wens>
arokux2: I forgot to enable Thumb mode support.
<arokux2>
forgot? as if you knew :p
<arokux2>
wens, btw, just got usb working and go to bed.
<arokux2>
wens, lots of polishing ahead of me.
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 245 seconds]
arokux2 has quit [Remote host closed the connection]
Black_Horseman has quit [Ping timeout: 240 seconds]
Black_Horseman has joined #linux-sunxi
gzamboni has joined #linux-sunxi
pacopad has joined #linux-sunxi
pacopad has quit [Quit: pacopad]
leviathanch has quit [Quit: No Ping reply in 180 seconds.]
Turl has quit [Ping timeout: 245 seconds]
ynezz has quit [Ping timeout: 264 seconds]
Turl has joined #linux-sunxi
ynezz has joined #linux-sunxi
fldc has quit [Quit: quit]
stekern has quit [Ping timeout: 245 seconds]
stekern has joined #linux-sunxi
atiti has quit [Ping timeout: 256 seconds]
fldc has joined #linux-sunxi
<oliv3r>
arokux: what did you have to fix?
ykchavan has joined #linux-sunxi
ykchavan has quit [Ping timeout: 260 seconds]
<mnemoc>
moin
panda84kde has joined #linux-sunxi
<gzamboni>
moin
<gzamboni>
sup mnemoc are u in the north of germany ?
<gzamboni>
this is what they were used to tell me to say hi
n01 has joined #linux-sunxi
<gzamboni>
i have been away from the sunxi stuff for a while, now im back. I followed a little the ML and some discussions
<gzamboni>
a lot of work have been done
akaizen has joined #linux-sunxi
akaizen_ has quit [Read error: Connection reset by peer]
akaizen_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
rellla has joined #linux-sunxi
TheViking has joined #linux-sunxi
<oliv3r>
gzamboni: welcome back!
CoSM0 has joined #linux-sunxi
<TheViking>
Welp, one little thing wrong and everything changes...
<mnemoc>
gzamboni: moin comes indeed from the north of germany, but it's pretty common on irc as a timezone-less greeting
<TheViking>
I have never bought a Serial-to-usb cable before. My cubieboard from miniand came with a Prolific PL2303 which doesn't appear to work. Could anyone recommend a good serial-to-USB cable?
<mnemoc>
TheViking: try swapping the Rx/Tx cables. be sure to connect Gnd, and NEVER connect VCC
<n01>
TheViking: FTDI
<mnemoc>
you can buy a dozen pl2303 for one tfdi :p
<gzamboni>
thanks oliv3r you all are doing a great job
<TheViking>
mnemoc: Check, Check, and Check. Also attempted to use another cubie I've had before + my modem (which I've put headers on but used a homemade db9 adapter) + a wm8505 board. Also tested on 2 linux boxes and windows 7.
<TheViking>
n01: Thank you for the tip.
<gzamboni>
so from what i see A20 is well supported and everything is advancing in the mainline
<wens>
The TTL pinout on the wiki is wrong. reverse TX and RX fixes it.
<JohnDoe_71Rus>
pins write on the board
<TheViking>
Wens: I did try that, but only with one cubieboard, nothing else. I'll check that on something else, like my wm8505
<wens>
JohnDoe_71Rus: maybe the color for the included cable is wrong. I don't have anything else to test it with.
<TheViking>
Nope. Unless it's really messed up (black ISN'T ground, for example) I have no clue. I can't test my modem as I'm currently using it :-P
<JohnDoe_71Rus>
There are success boot android with sd card? cubie2 A20
<TheViking>
Also, has anyone tested/had sucess with using f2fs as the rootfs type on a 1GB A10 cubieboard?
n01 has quit [Ping timeout: 246 seconds]
<TheViking>
mnemoc: You weren't kidding. Sparkfun has a 5-pin cable for $20+ after shipping
<TheViking>
Plus I would have to solder on individual headers.
<tm512>
I really like having a computer that can't display anything on screen
<tm512>
I've got an image that should be doing 1080p for the cubieboard, but I get nothing on screen
<ynezz>
just try to switch on the screen
<tm512>
I've turned the screen on and off, nothing
<JohnDoe_71Rus>
tm512: use hdmi-dvi converter?
<tm512>
I have to use an hdmi to dvi cable
<tm512>
what do I do?
<JohnDoe_71Rus>
some trouble with some converter and tv/monitors
<arokux2>
however I find it odd, since pdev->dev.coherent_dma_mask is used and then overwritten. can anybody explain?
<oliv3r>
arokux2: looking at the original code :p
<arokux2>
oliv3r, well, the point is original code still worked without it! either there was a change in mainline or Allwinner guys have done something somewhere else.
<TheViking>
If I wired an 1MAh LED to the TX-Ground on a cubie, it would blink from activity, correct?
<TheViking>
I need a way to test to see if my board's serial is bad :-<
<oliv3r>
TheViking: 1 mA is a little weak; but yeah in theory you can wire 2 leds on the tx and rx lines
<oliv3r>
just remember there will be a voltage drop due to the led
tinti has joined #linux-sunxi
<arokux2>
oliv3r, so desperatly looking for "differences" I've added this dma shit and it worked..
<oliv3r>
good job :)
<arokux2>
oliv3r, ok, I've missed something original code also refuses to work somehow.
<hno>
yes, that seems to be the origin. See 3b9561e9d9b88eca9d4ed6aab025dec2eeeed501
<hno>
And...
<hno>
+ * use reasonable defaults so platforms don't have to provide these.
<hno>
+ * with DT probing on ARM, none of these are set.
<hno>
whatever that means.
<arokux2>
hno, how have you found it?
<vinifr>
mripard, ping
<hno>
arokux2, git annotate
<hno>
anyway, copying what all the other drivers does is better than trying to fix it at one place.
<arokux2>
hno, trying to fit it in one place? you mean invent something?
<hno>
arokux, no, i said copy, don't try fixing it.
<hno>
and maybe ask Arnd Bergmann why it was written like that and now copied everywhere. But should not matter for ARM.
<hno>
Ah, I see it now. There is an &
* hno
feels silly.
<hno>
order does not matter.
tzafrir has joined #linux-sunxi
<arokux2>
hno, yep, you are right! good eyes !
ykchavan has joined #linux-sunxi
hansg has joined #linux-sunxi
arokux2 has quit [Remote host closed the connection]
Soru has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
<arokux>
are we supposed to work closer with arm-netbook guys?
<mnemoc>
most of us are there too, and lkcl_ (eoma guy) is here
<mnemoc>
actually this channel is an spinoff of arm-netbook. but eoma68 is SoC independent and this one is allwinner-centric
<arokux>
yes, this is known to me. however they seem to have there (confusing) wiki and spent lots of time to compile the kernel, which with our tutorials is straight forward.
<arokux>
maybe we can make sure our sunxi-3.4 runs well on eoma and have branch for it?
<mnemoc>
there is zero need for per board branches
<mnemoc>
all board stuff is in the script.bin or .dts file, kernel is the same
ykchavan has quit [Ping timeout: 256 seconds]
<arokux>
mnemoc, than it is even better. I've just thought for in case they were needed
<arokux>
their kernel is ancient with -C to standby and gcc-4.4 hard coded into the tree!
hipboi_ has quit [Remote host closed the connection]
iamfrankenstein has joined #linux-sunxi
TheViking has quit [Quit: Page closed]
pacopad has joined #linux-sunxi
<atsampson>
is there a mailing list for the allwinner-dev-team Android stuff, or is it just linux-sunxi?
<atsampson>
there seem to be about a dozen different CyanogenMod-to-A10 ports, so I'm wondering if there's somewhere central to discuss them...
<arokux>
hno, you have the EOMA hardware, maybe you can make sure everything needed is in our repos (fex, u-boot support) and check the latest and greatest sunxi-3.4 work fine with it?
<libv>
arokux: a reply to the pengpod news at phoronix
<oliv3r>
libv: huh what?
<arokux>
libv, you see this is also politics :p
<arokux>
so how will they do graphics on linux?
<libv>
arokux: yes, it is, but this sort of powerplay is approached from another angle
Seppoz has joined #linux-sunxi
<libv>
arokux: because a20 is more likely to be open, it is more popular and has a bigger part of the community behind it
<oliv3r>
atsampson: which is ironic, because there only needs to be 1 A10 CM tree, and a couple of HW packs with the various script.bins, dts and special modules (though that could even be done in 1 big pack)
<libv>
arokux: there is no one instance that tries to enforce this, this is a general preference and noone is actively excluding a31
<oliv3r>
atsampson: the problem is, these kids tend to have hardly any clue what they are doing, so it's a mish mash of crap
<oliv3r>
arokux: hno has sent his emoa sample back i belive
<arokux>
libv, at the last one you are of course right. we won't prevent nobody from adding support for A31
<arokux>
oliv3r, really?? why?!
<oliv3r>
libv: ohh phoronix; yep :D
<oliv3r>
arokux: i think there where only 5 initial samples; and hno got 1 to add support
<arokux>
oliv3r, and then he sent it back?
<oliv3r>
libv: well it's true isn't it ;)
<libv>
oliv3r: you're bang on, there is just one "but" in your statement
shineworld has left #linux-sunxi ["Sto andando via"]
<libv>
oliv3r: and that is with that bastard lima developer finally delivering something useful
<oliv3r>
libv: :p :D
<oliv3r>
well i'm proud of the work you do!
<oliv3r>
i just didn't wanna lay it too thick ontop :)
hansg has quit [Quit: Leaving]
<libv>
now if only i hadn't wasted as much time on complaining about intel, i would've started cleanup already on the mesa driver
<arokux>
oliv3r, what is the nickname of that pengpod guy you know? we may want to ask him to post their gpl stuff, or?
Black_Horseman has quit [Remote host closed the connection]
<oliv3r>
arokux: drachensun
<oliv3r>
arokux: he's here regularly and has posted patches
<arokux>
oliv3r, ah, ok
<oliv3r>
though with a31 I guess that'll be mostly the AW supplied kernel
<arokux>
oliv3r, it is not in our sunxi?
<oliv3r>
we have very little a31 support, if any
<oliv3r>
and personally; i don't care much for a31 due to powervr
<oliv3r>
arokux: having said that, chances are, that a40 or whatever their next soc is, will look a lot like a31 (but hopefully with mali). in that case, i'll probably work more on u-boot for that, as I started with patches for it
<arokux>
oliv3r, ok, I see, but why not push them to release the code using the momentum? we may not work on it actively, but..
<oliv3r>
arokux: we have the lychee code
* mnemoc
has a lychee (chinese) restaurant next to his current apartment
<arokux>
hm.. and that is the last kernel from sdk 3.2?
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<Tsvetan>
¨There’s even some support for 3D graphics and HD video playback in Linux using the PowerVR graphics chip.¨
<Tsvetan>
libv ^:-)
deasy has joined #linux-sunxi
<oliv3r>
arokux: nobody knows (or cares)
<arokux>
oliv3r, as said, we can use momentum to get sources, what ever happens to them later..
<Tsvetan>
Since the developers already had experience working with Allwinner chips ...
<Tsvetan>
do you know who these developers are? Pengpods are just Chinese manufactured tablets re-sold under this name at double price
<arokux>
:)
<arokux>
More importantly, PengPod’s developers says Allwinner has been pretty good about providing the source code and documentation needed to get GNU/Linux working on devices with the quad-core processor.
<arokux>
oliv3r, and then you say nobody knows they are the most recent onse
<arokux>
ones*
<oliv3r>
nobody cares
<oliv3r>
:)
<oliv3r>
if it's 3.3 or 3.4
<oliv3r>
we have 'something' we use it for reference sometimes
<arokux>
ok, I care to get sources, because it will be easy to do.
<mnemoc>
arokux: the task is to turn those sources into something that doesn't make you puke and the device crash
<oliv3r>
and for a31, nobody is really motivated to do that, due to the powerVR :)
<oliv3r>
we prefer to put our resources into a10 and a20; as we can have a fully opensource SoC with those
<mnemoc>
unify the different sunNi branches and SDKs, and then prepare for mainlining
<mnemoc>
oliv3r: libhybris probably
<oliv3r>
and resources are thin as they are
<arokux>
mnemoc, I see, I'm not interested on working on them, but it is nice to have them. penpods will have a lot of attention in media and the last thing they won't to be is gpl-violators :p
<oliv3r>
mnemoc: libhybris for?
<mnemoc>
pvr
<oliv3r>
*puke*
<mripard>
vinifr: pong
<arokux>
again: I do not want to force you oliv3r to work on A31, but to get the last sources, yes, I want them.
<mnemoc>
arokux: and DHL lost a ~1000E box of these tablets for drachensun I sent him :(
<mnemoc>
they were supposed to be giveaways for the press conference
<mnemoc>
f* dhl
<arokux>
dhl??? have they paid back?
<vinifr>
mripard, why to use val2 rather than val?
<oliv3r>
arokux: then go get them tiger! :)
<oliv3r>
arokux: i think the 3.2 SDK has been released for a31? or not yet?
<arokux>
oliv3r, yes
<oliv3r>
mnemoc: you sent what?
<mnemoc>
oliv3r: I was drachensun's proxy, receive the box and forwarded it to him
<oliv3r>
mnemoc: ohh; well atleast with DHL its inssured?
<libv>
mnemoc: dhl is truly the better of package deliverers in germany
<mnemoc>
but dhl delivered it back in some random location here in berlin, instead of orlando
<libv>
mnemoc: you will learn to frown at hermes, and dispise dpd
<oliv3r>
UPS :p
<oliv3r>
Oh, PostNL! :)
<libv>
ups is pretty rare here, but yes, also active here
<mnemoc>
libv: I shipped the box from Potsdam to Orlando, and it was delivered "somewhere" in Berlin the next day
<arokux>
oliv3r, well, there are devices with 3.2 SDK, does it mean it was released?
<mnemoc>
libv: and they only know the name in the signature... not even the address where they delivered it
<libv>
:(
<libv>
they should know that quite well though, things are tracked quite thoroughly
<pacopad>
Hi could , when do you think A20 cerdarx could be usable ?
<pacopad>
is it a question of days , week , mont ?
<libv>
and when they know the time, they should know which part of which street it was delivered to, as the dhl guys in germany really do just a few streets in a big city
<arokux>
pacopad, the classical is "when it is ready"
<pacopad>
:)
<mripard>
vinifr: scale * voltage_raw is the value in volts
<arokux>
pacopad, help us!
<mripard>
from your code, it looks like you compute micro volts
<pacopad>
why not
<pacopad>
what must i do ?
<mripard>
so it should be in val2 (the micro part), and not val (the int part)
<arokux>
pacopad, talk to ssvb and wingrime, they are on that part
<oliv3r>
arokux: ask drachensun if he received the sdk for a31 and if its shared; wouldn't suprise me if it where
<oliv3r>
mnemoc: well atleast it's insured
<libv>
mnemoc: but as is usual with these things, you get a civil servant who cannot be bothered at the other end of the line
<arokux>
oliv3r, what do you mean by "if it is shared"?
<oliv3r>
arokux: not everybody wants to host it :p
<arokux>
oliv3r, "if it is share" = if it is available for download?
<vinifr>
mripard, ok
<mnemoc>
libv: and the online system already forgot about the package....
<oliv3r>
allwinner sends their SDK's to distributors etc, they don't share it themselves; so until one of them re-shares it, it's 'hidden'. Tsvetan always shares everything he gets for example :)
<oliv3r>
pacopad: expect years, maybe months; not sooner
<Tsvetan>
we do not buy A31 chips so I have no ground to ask AW for SDK
<mnemoc>
libv: awfully depressing
<Tsvetan>
but they never refused to send SDK for the chips we buy
<mnemoc>
libv: "the procedures are perfect".... problem is, people performs them....
<oliv3r>
Tsvetan: good! :D
<oliv3r>
mnemoc: did you get your money back?
<Tsvetan>
arokux but we have no board for them :)
<mnemoc>
oliv3r: drachensun's money. but afaik they might still recover the box
<oliv3r>
mnemoc: ok, well maybe he'll get money and a box :p
<oliv3r>
mnemoc: why where you the proxy?
<vinifr>
mripard, What other values should be removed from dt? I left startup_time, sample_hold_en
<mnemoc>
oliv3r: i seems the resellers only shipped to .de
<mnemoc>
weird start :|
<mnemoc>
via DHL
<oliv3r>
mnemoc: ah ok fiar nuff
<arokux>
Tsvetan, I see
<mnemoc>
and the DHL office where I received it didn't accept shippings.... so I had to schedule a pick up from work instead of forwarding it directly....
<Tsvetan>
mnemoc if you had tick the insurance checkbox you would save these toubles :)
* mnemoc
hates couriers
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
Soru has quit [Read error: No route to host]
soul has joined #linux-sunxi
soul is now known as Guest33091
<mripard>
vinifr: basically, all of them but the vref
<vinifr>
mripard, Ok, so sunxi_adc_probe_dt() is no longer necessary. Why did you use at91_adc_probe_dt in at91_adc?
wingrime has joined #linux-sunxi
<mripard>
because at91 can be loaded either by DT or old-style board files
<arokux>
oliv3r, mouse/keyboard can be tested in DE really, or? or maybe through some debugfs stuff?
<mripard>
arokux: just so I know, you seem to have added a Booting Kernel quickly section to mainline
<arokux>
mripard, yes I did, something wrong with it?
<mripard>
I don't really get the point of it, since you're duplicating the content of your A10 netboot page
<mripard>
(which is great)
<mripard>
plus saying some stuff about script.bin
<arokux>
mripard, mm.. you are right, was asking myself the same. I've just thought to give a quick recipe where as the other page is in-depth
<wens>
n01: I thought they went over this like a month ago?
<n01>
wens: nope :)
<mripard>
arokux: I think adding a reference to your NFS page
<mripard>
which is already pretty great
<mripard>
and covers everything needed
<arokux>
mripard, as to script.bin: I boot both kernels, just to see the differences in registers etc. that is why i thought it could be useful here to add here info as how to boot different kernels (mainline and 3.4) with the same bootloader using a few keystrokes.
<oliv3r>
arokux: well keyboard you just plug in and press a button :p
<oliv3r>
arokux: mouse, cat /dev/input/mouse and move your mouse, or get gpm
hipboi has joined #linux-sunxi
<arokux>
mripard, now I have NFS setup, works great. I get u-boot shell and run either "run ml" or "run aw" and the proper kernel boots
<arokux>
oliv3r, hm.. kernel will say something if pressing a button?
<oliv3r>
arokux: ahh wait, we have serial console, and your keyboard presses won't show up there ;)
<oliv3r>
well try /dev/input/keyboard
<arokux>
mripard, so those were my thoughts.
<arokux>
oliv3r, ok, will try..
specing has quit [Ping timeout: 264 seconds]
robertjw has left #linux-sunxi [#linux-sunxi]
<mripard>
arokux: I completely understand the point of it
<mripard>
I just don't get how it's related to this page.
specing has joined #linux-sunxi
<mripard>
(and scripts.bin has *nothing* to do here, but it's a minor issue)
<arokux>
mripard, so I thought it would be nice help for people who hack on mainline, a quick recipe how to boot both kernels.
<arokux>
mripard, maybe they aren't even aware of the fact one u-boot can handle both.
<wens>
f
wingrime has quit [Ping timeout: 264 seconds]
VicenteH has quit [Read error: Operation timed out]
<arokux>
mripard, if you think it is better there could be a page titled "Best practices while hacking on mainline"
rz2k has quit []
wingrime has joined #linux-sunxi
<oliv3r>
arokux: quick howto's are good I suppose, but most devs should figure out most stuff or hang out here anyway to ask ;)
<arokux>
oliv3r, they ask, then they mistype and ask again and loose there time. so why not copy-paste working commands?
<oliv3r>
arokux: that's easy howto's :p
<arokux>
oliv3r, well for you too, but a newcomer (as me) to the embedded world wanting to help out with kernel will spent several days putting those command together. he could get frustrated in between and go away...
<oliv3r>
arokux: pf,f no endurance, not worth it ;)
<oliv3r>
arokux: but i've only started here not too long ago ;)
<arokux>
oliv3r, you seem to have a better background
<arokux>
but what about we are talking now? recipes are useful. you claim they are for noobs, I agree. they can be moved somewhere into category for noobs, so go ahead :)
\\Mr_C\\ has joined #linux-sunxi
vinifr has quit [Ping timeout: 240 seconds]
<atsampson>
oliv3r: (re Android) that was kind of the impression I'd got ;-)
<atsampson>
but even someone who's done a messy port has still figured out what modules a device needs, etc., so it'd be useful to collect that somewhere for a proper port
<libv>
mnemoc: would it be possible to house mesa packages for debian and ubuntu at sunxi.org?
<oliv3r>
atsampson: i think someone started that a few weeks ago? was it you? :p
<libv>
these packages would differ from the standard ones in that they have (build system) changes so that the lima mesa driver can be built externally
<oliv3r>
atsampson: ideally, we'd have a generic allwinner CM/replicant tree, where you can add the kernel and pretty much all related kernels
<oliv3r>
then you need per device script.bin (for now, dt later) and per device u-boot (for now) so HW-packs or seperated tree's are required for that
<arokux>
is guy with wiki nickname Sssemil here?
<libv>
i am currently happily working on 9.0 mesa packages, but i think that i will also build 8.0 (which is the oldest of the currently used mesa versions, ubuntu lucid LTS and debian stable)
<atsampson>
oliv3r: that's more or less what's already there in the allwinner-dev-team bits (which I think is Turl's?)
\\Mr_C\\ has quit [Ping timeout: 245 seconds]
<oliv3r>
libv: exciting stuffs! and i'm sure its no problem! :D
<oliv3r>
atsampson: turl is/was part of it yeah
<Turl>
libv: there should be no trouble hosting that
<oliv3r>
atsampson: one of the things these people do, with their 'per device tree's is add blob drivers specific for their hardware
<arokux>
o_O what is Allwinner Dev Team?!
<Turl>
DHL lost mnemoc's package? I'm highly surprised</irony>
<Turl>
arokux: github org hosting android repos
<libv>
these would all be drop in replacements for existing packages, and would not require anyone to change anything else to their systems
<arokux>
libv, this is nice, people always want something like this
<oliv3r>
rellla: you should try experimenting XBMC with these drivers ;0
<arokux>
I hope android will die eventually just after normal distros could run fast enough
<oliv3r>
arokux: there's a ton of apps though; don't forget that, you'd need some sort of support for the apps
<oliv3r>
arokux: i imagine some sort of KVM sandbox taht runs the app
<Turl>
arokux: I hope it doesn't, it's great for mobile :)
<atsampson>
arokux: it's the lack of touch-friendly interfaces that's the problem at the moment -- my tablet runs Debian perfectly well but I can't actually do much without a keyboard on it ;)
<libv>
arokux: just you wait, i will get a lot of flack for doing things this way
<libv>
arokux: mostly from people who are not actively involved or are dependent on this work though
<libv>
for average joe user, it will be a breath of fresh air, that's true
<oliv3r>
libv: put on your flack-jacket
<rellla>
oliv3r: i'm very curious, how all the efforts made these days will work together
<oliv3r>
we'll be ready for em
<oliv3r>
rellla: well blobs cause headaches :p so we'll see
<arokux>
atsampson, agree 100%, kde people were working on it. i think as arm cpu will be more powerful we'll see a lot of opensource projects
<rellla>
oliv3r: atm i'm idling and reading here and enjoy my cubie2-logitechmediaserver. no doubt. it's very much faster than good old dockstar :p
<arokux>
libv, do you have something like FeatureMatrix for mali? curious.
<libv>
arokux: it's called limare/tests :p
<arokux>
rellla, what is there from logitech?
<rellla>
arokux ?
<arokux>
rellla, "cubie2-logitechmediaserver"
<arokux>
libv, I do not know if you remember, but I've contacted you a year ago or so to get a small task
<rellla>
arokux: formerly known as squeezebox server. for my squeezebox. since there is no prebuilt package for armhf, i built it myself and enjoy it ;)
<libv>
arokux: when an initial mesa driver is out, there will finally be a situation where all sort of small tasks will be possible
<oliv3r>
rellla: logitechmediaserver? what's that :)
<oliv3r>
the guy on the olimex forums was complaining he couldn't get it to work at all
<oliv3r>
arokux: doubt it, the AXP is an entire microcontroller, with its own IRQ's etc
<ssvb>
wingrime: adding lots of debugging prints and comparing the logs from a13 and a10 could probably help
<ssvb>
wingrime: but probably the traces of the binary blobs would be also needed to see what exactly is different
<ssvb>
wingrime: have you tried it on a20?
<arokux>
oliv3r, so what is the next driver? ;)
<wens>
arokux: it's a beast
<arokux>
(that would be nice to have)
<oliv3r>
well i only have to do a few things with PWM but have to test it
<oliv3r>
arokux: ohh, erm
<wens>
I was thinking about the crypto engine. wonder if it's actually doable with only the user manual and no reference
<oliv3r>
arokux: SmartCardReader? :p
<oliv3r>
gmac driver port from the 3.4 SDK?
<oliv3r>
emac driver could use some cleaning up; but getting the gmac driver to work with the 100mbit PHY would be cool (should support it)
<oliv3r>
that way, the emac driver could be merged/removed witht he gmac driver
<arokux>
oliv3r, how is gmac different from emac?
<wens>
oliv3r: i thought they used different pins?
<oliv3r>
arokux: emac only does MII; gmac does GMII, RMII and RGMII
<oliv3r>
wens: different mux on the same pins
<oliv3r>
arokux: it's completly different IP
<wens>
oliv3r: oh you're right
<arokux>
oliv3r, in which SoC does gmac reside?
<oliv3r>
arokux: but emac IP is known to have some hardware bugs (check mailing list) so using the gmac driver for both would be a boon. emac is still needed for A10 though
<oliv3r>
arokux: only in A20
<oliv3r>
arokux: a20, a31(s) and persumably in a23
<arokux>
I see.
<arokux>
oliv3r, what do you mean by SmartCardReader?
<arokux>
mmc?
<oliv3r>
mmc != scr
<arokux>
i'm not familiar with terminology
<wens>
arokux: there's a smart card reader embedded in the soc
<arokux>
wens, any device that exposes it to the world?
<oliv3r>
arokux: Smart Card Reader
<wingrime>
ssvb: still not on a20
<wens>
arokux: think ic credit cards
<oliv3r>
arokux: SIM, Creditcard, pay-tv?
<arokux>
oliv3r, ok, I see now. but why should this be interesting for mainlining? we do not have hardware even
<oliv3r>
arokux: just a thought ;)
<wingrime>
ssvb: also, good news for you
<wingrime>
ssvb: do you remeber screen shake on 1920x... screen
<arokux>
oliv3r, hm.. what about NAND?
<arokux>
or mtd.. again not familiar with terminology
<oliv3r>
arokux: the MTD driver needs some love yeah, OOB still doesn't work? but that's quite intensive to get work
<arokux>
oliv3r, OOB?
<oliv3r>
arokux: well we have 'libnand' driver, which is allwwinners own nand driver, it has been ported to MTD
<oliv3r>
arokux: out of bound i think i twas; i don't know much about nand :)
<Turl>
arokux: no, it's based on clk-next+the sunxi dt pull
<arokux>
Turl, I'll add usbclk module soon what are your recommendations as against which branch should I diff
<ssvb>
wingrime: the kernel might be tweaked to automatically (and transparently to userspace) enable scaler mode for the desktop layer if the scaler is not used by anything else
<wens>
found the gmac driver
<ssvb>
wingrime: manually flipping the scaler mode or adding it to fex is a big pain for the developers and/or users
<Turl>
arokux: I'll rebase my branch again to torvalds/master once the sunxi pulls reach it
<Turl>
I think there's 1 or 2 missing still
<arokux>
Turl, so you say I better wait with diffing?
Guest33091 has quit [Quit: No Ping reply in 180 seconds.]
soul has joined #linux-sunxi
<Turl>
arokux: as long as you are working on a1x you can work on top of my branch now and rebase later on
soul is now known as Guest91485
<arokux>
Turl, ok, thanks
<arokux>
I wonder how chinese guys could produce something faster than olimex, olimex beiing a much older company..
<wingrime>
ssvb: we can simply ignore fex option when we using a10 , or/and on some resolutions , on a13 we have only one, so, we must not use it
<Turl>
ssvb: I know the article is old, but I think HugeTLB and THP reached mainline already
<ssvb>
Turl: yes, I know :-)
<ssvb>
Turl: that's a good stuff
<arokux>
wingrime do you work on sata?
<wingrime>
arokux: I wait clocks settled to mainlne
fldc has quit [Quit: quit]
<Turl>
wingrime: when mainline settles I will make branch for you
<Turl>
wingrime: you can port sata on top of it on A1X/A20 hw
<arokux>
wingrime, Turl, only PLL6_SATA is needed or?
<Turl>
arokux: yes, that should be it
<wingrime>
arokux: also don't forget gate on
<arokux>
Turl, I've added PLL6 clock, because I needed it
<arokux>
I can post a patch tonight, just tell me against what branch should I diff
<wingrime>
hipboi: are you moved to rockchip ?
<Turl>
arokux: I already have pll6 implemented since quite a while ago
<hipboi>
wingrime, not moved to
<hipboi>
wingrime, both rockchips and sunxi
<Turl>
arokux: it was just needing a dt node for a20 for wingrime to work on which unsurprisingly ended up being a copy-paste of the A10 one
<wingrime>
jemk: I do trace, but I think many frames will be lost
<jemk>
wingrime: add --rate 0.05 if you have time, then no frames will get lost ;)
<pacopad>
@wingrime @jemk : hi guys , what's the way to test libvpau-sunxi on a20 ? i compiled it and tried with gnome-mplayer but i got a black screen :(
<jemk>
pacopad: can you test with pure mplayer, i don't know if the output handling is able to handle more complex output windows
<wingrime>
jemk: also ,please add workaroud for negatvie y
<arokux>
mripard, could we add usb hosts into merging plan for 3.13?
<arokux>
oliv3r, what is the point of merging SID driver without device tree?
<mripard>
hmmm, I agree with the hack screaming part, but it's kind of common
<mripard>
I know that mxs does this at least
<mripard>
I'd say that the point of a framework is to be extended :)
<mripard>
but you can always submit a v1 with this, and see how mturquette yells :)
<Turl>
I guess :)
<arokux>
Turl, do you think clocks = <&tegra_car TEGRA30_CLK_DSIA> is better than clocks = <&tegra_car 478>?
<Turl>
arokux: it's easier to the eye I guess
<mripard>
arokux: I don't.
<Turl>
last I checked mripard wasn't fond of .h'ing the dtsi, I guess he still isn't :)
<arokux>
hiding? what hiding?
<mripard>
he didn't say hiding :)
<arokux>
mripard, I just want a number at one place only, why this is bad?
<tm512>
where do I find the mkimage command?
<arokux>
tm512, in one of the packages supplied by your distro
<mripard>
arokux: several things come to my mind. When you debug the DT, you then have to have open: the datasheet, the DT, and the *right* header included
<mripard>
then, you'll have one header for one SoC, because you said it yourself, it's just *one* value
<mripard>
so there's not even the de-duplicating argument
<mripard>
to me, it just adds code to obfuscate stuff
<arokux>
mripard, ok
<arokux>
dt is bad anyway :(
<mripard>
I disagree here as well, but anyway :)
<Turl>
mripard: something broke my hacked-in cpufreq meh :(
<arokux>
n
ZetaNeta has joined #linux-sunxi
<arokux>
Turl, what is the point of having usb-ehci here? compatible = "ti,ehci-omap", "usb-ehci";
<Turl>
mripard: ^
<Turl>
arokux: probably because it's ehci-compatible :)
ganbold_ has quit [Remote host closed the connection]
<arokux>
Turl, will it have some consequences?
<Turl>
mripard: ^
<Turl>
probably for drivers looking for the parent ehci controller or something, but I'm guessing here
vinifr has joined #linux-sunxi
<mripard>
nah, it will first try to look for a "ti,ehci-omap" driver (and specific driver for that IP), and then fall back to the generic driver.
<arokux>
mripard, ok, so no use for us as I cannot imagine the generic driver too know how to set all the clocks
<mripard>
it's useful for example when you have an IP that is compatible with a given driver, but yet has some additionnal features
<mripard>
that way, you'll fall back on the driver implementing the common denominator features if you don't have support for that precise IP
<arokux>
mripard, makes sense, thanks for info
shineworld has joined #linux-sunxi
<mripard>
you know better the ehci stuff, but yeah, you're probably right :)
<arokux2>
Turl, who has access to the sunxi.org? can he create a folder named sdk and put all the sdks in it?
<drachensun>
I had some work on in a merged community kernel that would run A31 as well
<drachensun>
but it was going slowly and I think I ended up losing the files to a faulty drive
<Turl>
mnemoc: ^
<mnemoc>
Turl too
<arokux2>
drachensun, what kind of support does the kernel has for GPU?
<oliv3r>
i think we have the previous sdk allready
<arokux2>
drachensun, "There’s even some support for 3D graphics and HD video playback in Linux using the PowerVR graphics chip."
<arokux2>
drachensun, details are interesting!
<drachensun>
libhybris is how I am doing that
<Turl>
mnemoc: I can, but I don't get to decide if we can host multi-GB files and eat all the bw :)
vinifr_ has joined #linux-sunxi
<drachensun>
I got the mer/nemobile wayland QT demo thing running
<drachensun>
so its standard linux GPU stuff
<drachensun>
a kernel module we can build
<drachensun>
a giant blob user space library that does the real work
<arokux2>
drachensun, ok, so as usual
<drachensun>
yup
<oliv3r>
mnemoc: what is the builder-v4 renamed to?
<arokux2>
oliv3r, you seem to know devm_ stuff. should it be released? i.e. is it safe to do a = devm_get_xxx; two times?
vinifr has quit [Ping timeout: 264 seconds]
<oliv3r>
arokux2: fromw hat i learned from mripard, it will get cleaned up after module unload
<arokux2>
oliv3r, magic.
<oliv3r>
that's the point of the devm_ functions
<arokux2>
oliv3r, but if the module sits there and we have issued let say 5 devm_get_xxx and failed on the last?
<arokux2>
oliv3r, should it be ignored?
<oliv3r>
good question, i would expect the failure will cause the call to clean itself up; if you mean the entire driver fails to load; then all 4 will get cleaned up (as the 5th failed)
<arokux2>
oliv3r, I see. makes sense.
jemk has joined #linux-sunxi
<oliv3r>
Turl: mnemoc downloading the lychee-3.2 thing @ 32k; if that's the kernel, not sure if we want to bother with the full android thing
<Turl>
oliv3r: :p
<oliv3r>
Turl: unless you suggest otherwise
<oliv3r>
once it's in /srv/download, you can decide what to do with it ;)
<Turl>
oliv3r: I can handle the android DL if you wish
<oliv3r>
i don't know if we really need/want it
<Turl>
oliv3r: it can't hurt can it? :)
<oliv3r>
waste of space :p
<oliv3r>
but i suppose :)
<arokux2>
I need to know the number of the controller: 1 or 2, should I decide on its address in probe?
<oliv3r>
arokux2: USB?
<arokux2>
oliv3r, yes
<arokux2>
oliv3r, forget
<arokux2>
oliv3r, no, do not forget :D
<Turl>
arokux2: I think you need to have a binding like this, sec
<arokux2>
Turl, what about sun4i-xxx-clk -> sunxi-xxx-clk in sunxi-clk?
<oliv3r>
arokux2: same goes for sun4i-emac -> sunxi-emac
<arokux2>
hmm.. don't we want to have dtsi for AXXx? so that interrupts can be overwritten?
[1]MacMickey has joined #linux-sunxi
<oliv3r>
arokux2: what do you mean/
<arokux2>
for example, ehci is the same on a10 and a20 however interrrupt + 30
MacMickey has quit [Ping timeout: 246 seconds]
[1]MacMickey is now known as MacMickey
<oliv3r>
arokux2: i don't think you are supposed to use dt like that; sun4i-a10.dtsi is your core dts, if sun7i-a20 is ALMOST the same, that's great, but you still have to copy/paste it :)
<oliv3r>
arokux2: and isn't it + 32?
<arokux2>
ok, +32)
<arokux2>
well that is why dtsi are there, or?
<arokux2>
usb 2-1: Product: 802.11n WLAN Adapter yay!
<Turl>
arokux2: no, compatible strings indicate the first IP that's compatible; in most cases sun4i
jemk has quit [Remote host closed the connection]
jemk has joined #linux-sunxi
<oliv3r>
Turl: we don't use regmap at all with our dts's do we?
<Turl>
yay, travelling salesman movie is available :P
<Turl>
oliv3r: not afaik
<oliv3r>
shouldn't we?
<Turl>
oliv3r: what's regmap? :)
<arokux2>
Turl, compatibles can be overwritten too?
<oliv3r>
i'm looking at the imx and i see they use 'remap_update_bits' to change a whole bunch of bits in one go
<oliv3r>
so i find 'syscon_regmap_lookup_by_compatible("fsl,imx6q-iomuxc-gpr");
<oliv3r>
doesn't that look atleast somewhat pretty?
<oliv3r>
unfortunatly i can't figure out it works, they use some syscon/mfd thing, and i can't really find tthose registers in the dt, I did find a .h file with all those registers in it
<oliv3r>
confusing
<arokux2>
Turl, C as KR's C?
<oliv3r>
K&R is the only C
<arokux2>
another noob question. there is alway problem with ioremap... should this func be called only once with enough size and the returned value used everywhere or can I call this func as often as i want for the same physical location?
<Turl>
oliv3r: well, sunxi-clk itself is buggy I think
<Turl>
there's a draft tag though I pushed the other day
<oliv3r>
08-21 is latest?
<Turl>
oliv3r: I'll clean it up this afternoon and update sunxi-clk
<Turl>
oliv3r: no, it's sunxi-clk-for-someone I think
<Turl>
oliv3r: do you need it for something? can you wait till later today? :)
<oliv3r>
i don't need it at all; allready found what I needed in your dts :)
<oliv3r>
some R&R first anyway, then bed; maybe some backlog reading in bed
atiti has joined #linux-sunxi
<arokux2>
you do not use ioremap? :p
shineworld has quit [Quit: Leaving]
<Turl>
"With a new fingerprint identity sensor, 64-bit A7 chip, and an even more impressive iSight camera, iPhone 5s is forward thinking. "
<Turl>
wonder if that's ARM64
wingrime has joined #linux-sunxi
<wingrime>
Tsvetan: ping
<buZz>
i doubt its ARMv8
pacopad has joined #linux-sunxi
<wingrime>
Tsvetan: I viewed your a13-som module preview
<wingrime>
Tsvetan: it have only once ddr3 ic
<oliv3r>
Turl: quite likly, '64-bit' kinda gives it away
<wingrime>
Tsvetan: but I heard that mali can't be runed with one ic (or 256?)
<wingrime>
oliv3r: ^
<jemk>
wingrime: can you trace some frames (one would be enough) of another video with different size (not 1920x1080)
<jemk>
wingrime: i think a13 needs some extra buffers for some temporary things
<wingrime>
jemk: are you have some ideas?
<oliv3r>
wingrime: mali can run fine with 1 IC, mali only needs 16 mb or something? so yeah it'll be heavily restricted if mem bandwith is crap and/or lots is required, but there's lots of usecases where you need only very basic video output
<oliv3r>
wingrime: we think now that a23 is a13, so sun8i
<jemk>
wingrime: i'll prepare a patch for you to test, but for buffer size i need two different values to verify my calculations
<oliv3r>
mnemoc: btw, we've dumped sun3i in case you hadn't noticed :)
MacMickey has quit [Ping timeout: 246 seconds]
<pacopad>
@jemk @wingrime i got an error when i try libvdpau on A20 with fresh sunxifb and mplayer
<pacopad>
mplayer could not open the selected video_out
<oliv3r>
pacopad: afaik a20 is untested etc
<oliv3r>
wingrime: ^
<wingrime>
pacopad:
<wingrime>
pacopad: do "chmod 777 /dev/disp
<wingrime>
and
<wingrime>
"chmod 777 /dev/cedar_dev
<wingrime>
if I remeber correctly
<pacopad>
same thing :(
<oliv3r>
wingrime: did you get cedarX working on A20?
<wingrime>
oliv3r: not tested
<jemk>
pacopad: where exactly it fails, can you paste full output somewhere
<jemk>
oliv3r: but even a13 doesn't work, so expect a20 to not work too
<pacopad>
yep
<wingrime>
jemk: it _basicly_ decoes _something_ _strane_ in green and while lines
<wingrime>
lkcl_: have a10/a13 any problems when it used with only one ddr3 ic or 256 mb ?
<Turl>
wingrime: I think problem is software only, because allocation for mali is/was hardcoded to 512-64 .. 512 region
<wingrime>
Turl: maybe deeper,
<jemk>
wingrime: if you give me trace of other size i'll give you some patch ;)
<Turl>
hno: you can find towards the end, 'If bootloader dont set Virtual Offset register,Physical Count Timer is needed to replace Virtual Count Timer.'
<Turl>
and then does some ifdef, which will not work with multiplatform
<wingrime>
jemk: cedar driver need unification
<wingrime>
jemk: but now
<wingrime>
jemk: there differencs
<wingrime>
jemk: a13 driver will ignore you freq setup
<wingrime>
jemk: and have no flush_cache instructions
<wingrime>
Turl: pgprot_noncached
<wingrime>
Turl: any idea about this function?
<Turl>
wingrime: noncached mapping
<Turl>
there was thread about it today on lkml
<wingrime>
Turl: it will be slowdown ?
<Turl>
lakml
<Turl>
wingrime: direct to ram I think, no L1/L2
<jemk>
wingrime: sun3i mmapes non cached, so flush is not needed
<Turl>
so yes
<wingrime>
jemk: I looks find problem
<wingrime>
jemk: sun4i have this function commented
panda84kde has quit [Read error: Operation timed out]
<Turl>
wingrime: if you remove pgprot_noncached you will need to flush I think
<Turl>
wingrime: otherwise memory may be incoherent from cedar
<jemk>
wingrime: ok, you'll find the problem, i have to leave now
<wingrime>
jemk: not 100% sure
<wingrime>
jemk: I need rebuild kernel and test ida
<wingrime>
*idea
<Turl>
bbl
<wingrime>
Turl: cedar have flush ioctrl
<Turl>
wingrime: if noncached is read/written directly from memory on usually small chunks
<Turl>
wingrime: and cached is read to fill the cache
<arokux2>
hno ping
<Turl>
wingrime: I think that is where speed difference is
<wingrime>
Turl: marketing crap
<Turl>
wingrime: read thread I linked
<hno>
arokux2, i am here, just ask.
<arokux2>
hno, how is support for EOMA68-A20 in U-Boot and is there a fex for it?
<arokux2>
hno, forget the question, everything is there.
<hno>
Turl, what do the other look like?
<arokux2>
hno, I've got usb working :p
<hno>
mnemoc, ok. not sure we are interested in those to be honest. There is practically no sun3i devices around, and no interest in using them.
<hno>
arokux2, any more changes after the dma bits?
<wingrime>
jemk: no
<arokux2>
hno, hell, I've told you today in the morning, no it was the last bit. now all the devices get correctly enumerated. wlan adapter included.
<wingrime>
jemk: thats not fixes
<wingrime>
Turl: thats :(
<wingrime>
jemk: so, you have find some more differences
<hno>
arokux2, what would you think about looking into adding the needed glue on the USB host driver in u-boot?
jemk has quit [Remote host closed the connection]
<wingrime>
Turl: there is more one issue
<arokux2>
hno, can do it, but please do a little intro to me where and what..
<hno>
arokux2, u-boot do have ohci and ehci, but no sunxi glue at all. See drivers/usb/host/
<arokux2>
hno, there is even musb directory o_O
wingrime_ has quit [Ping timeout: 264 seconds]
<arokux2>
hno, seems to be easy to add usb support.
<arokux2>
u-boot even has support for device tree, maybe i'll be able to reuse some stuff
<arokux2>
hno, how fast is it needed?
<deasy>
arokux2, you have succeed to boot on usb how?
<arokux2>
deasy, it was already possible with FEL mode afaik
focus_it has joined #linux-sunxi
<arokux2>
I've added USB host support to mainline
<hno>
arokux2, there is two musb directories even.. musb and musb-new
<deasy>
yes i have read it on wiki but this is with some command
<deasy>
arokux2, :o you hacker
tzafrir has joined #linux-sunxi
<hno>
arokux2, there is no panic, but many people have asked for it over the years.
<arokux2>
over the YEARS?! :)
<deasy>
XD
<hno>
yes, years..
<arokux2>
hno, but afaik ppl can boot over usb, or?
<hno>
we can FEL boot via USB, but that's not quite the same.
<arokux2>
hno, what exactly will be possible to do with usb host in u-boot?
<hno>
with USB host in u-boot you will be able to boot from usb-storage with only u-boot in nand/sd/mmc/nor.
<arokux2>
nor?
<hno>
SPI NOR to be exact.
<hno>
no known devices using SPI NOR, so don't focus too much on that part..
<arokux2>
and FEL?
<deasy>
bad popolon, he makes them working more !
<hno>
FEL is different. With FEL the AW CPU becomes a special USB device, and another host can drive the boot process (load code into SRAM, execute, load code into DRAM, execute)
<hno>
FEL mode is all in the internal bootrom embedded within the Allwiner SoCs.
<arokux2>
hno, so with FEL another machine is needed, whereas with usb host only storage device?
<hno>
yes
<buZz>
will there someday be sata support for uboot?
<arokux2>
hno, interesting, did they develop FEL mode by themselves?!
<wingrime>
hno: have we any nor spi support in uboot?
<deasy>
buZz, calm :p let's them add usb support :p
<buZz>
just curious if its possible at all
<hno>
arokux2, I think so. It's not very complex and is using the same USB device code they have in meli.
<tm512>
so fedora booted correctly for me
nove has quit [Quit: nove]
<deasy>
fedora always boot correctly
* deasy
don't troll too much
<hno>
wingrime, no, as there is no known devices with SPI NOR flashes there is no devices to test on. But it's not difficult to add if needed. u-boot as such do have full SPI NOR support, only missing a sunxi SPI driver.
<tm512>
the system was almost unusably sluggish
<arokux2>
hno, why could be SPI NOR interesting?
<tm512>
time to build my own highly optimized system
<hno>
arokux2, it's not. I said to not focus too much on that part.
<wingrime>
hno: a13-som
<arokux2>
hno, just curious. why would ppl want to use it? as there are so many other ops.
<wingrime>
hno: 16mb nor for boot
<hno>
wingrime, yes.
mturquette has quit [Ping timeout: 264 seconds]
<hno>
arokux2, SPI NOR is primarily used when there is no room for a nand or emmc device as primary boot device.
vinifr_ has quit [Quit: Leaving]
<arokux2>
hno, no room? so SPI NOR device can have extra small dimensions?
<hno>
arokux, normal package is SOIC-8.
<arokux2>
hno, yes, very small.
<arokux2>
hno, and much less pins then nand!
<hno>
there is even smaller. UFDFPN8 package is 2x3 mm including contacts.
<arokux2>
cool
<hno>
but normal package is the SOIC-8 narrow one.
hansg has quit [Remote host closed the connection]
<hno>
Turl, ok, found them. But does not make sense. the ,0 one is physical, ,1 is virtual. They differ in a virtualized guest only. Not sure why the patch is needed.
<Turl>
hno: is it related to TZ?
focus_it has quit [Quit: Leaving]
Seppoz has quit [Ping timeout: 264 seconds]
<hno>
No.
<hno>
I suppose it's for giving virtual guests an idea on how long their "virtual CPU" have been powered on.
tinti has quit [Read error: Connection reset by peer]
<hno>
Guess the offset should be initiealized to 0 by someone to not confuse. Default value is unspecified.
tinti has joined #linux-sunxi
tinti has quit [Read error: Connection reset by peer]
wingrime has quit [Ping timeout: 260 seconds]
naobsd has quit [Quit: Page closed]
<tm512>
can G2D be used without proprietary blobs?
<Turl>
tm512: yes, sunxifb does
<arokux2>
hno, forgot to ask, does u-boot supports sata?
<tm512>
I'm trying to build a musl libc system, hopefully everything works
<ssvb>
tm512: G2D is a simple 2D engine, which is nicely documented
<arokux2>
tm512, what is G2G?
<tm512>
the fedora image was very, very slow
<ssvb>
tm512: just change the cpufreq governor to performance, it will become much faster
<ssvb>
tm512: also if you use 1920x1080 screen resolution, consider changing it to something more modest
<tm512>
if I don't use 1080p, then it doesn't display anything
VicenteH has joined #linux-sunxi
Guest91485 has quit [Ping timeout: 268 seconds]
<tm512>
and even if it worked, it would be blurry, from scaling
<ssvb>
well, pick your poison
<ssvb>
you can also reduce desktop color depth to 16 bits per pixel
<ssvb>
and if it doesn't display anything when you are trying to change the resolution, then you are probably doing it wrong
<tm512>
if I can build everything I want on top of musl, then that should also improve responsiveness
<tm512>
I just built a cross compiler that uses musl as libc