xxiao has quit [Remote host closed the connection]
mysteryname has quit [Ping timeout: 246 seconds]
pwhalen has quit [Read error: Connection reset by peer]
cheng has joined #arm-netbook
pwhalen has joined #arm-netbook
alcides has quit [Quit: There's no place like 127.0.0.1]
QingPei has joined #arm-netbook
dfletcher_ has joined #arm-netbook
dfletcher__ has quit [Ping timeout: 260 seconds]
mysteryname has joined #arm-netbook
<Turl>
libv: pretty long half an hour isn't it? :P
* Turl
is intrigued still
<Turl>
oh, a new branch :D
tekzilla has quit [Ping timeout: 252 seconds]
tekzilla has joined #arm-netbook
<bsdfox>
does anyone here get decent X performance?
<Turl>
I do
<Turl>
(on my desktop :P)
avernos has quit [Ping timeout: 246 seconds]
hipboi has joined #arm-netbook
spittnacid has joined #arm-netbook
<spittnacid>
Hello from us. May i ask a question?
<Turl>
don't ask to ask, just ask :)
<Turl>
hipboi: ping
<spittnacid>
I have a onda vx610w resistive touch and interested in putting linux on it. Hopefully that runs at a decent speed and takes advantage of hardware acceleration. Is this possible?
<Turl>
spittnacid: what's the cpu on it, A10?
<spittnacid>
A10 mali400
<Turl>
yeah, you can probably run linux on it rather easily
<spittnacid>
I want to run it natively, no more android:-)
<Turl>
using GLES and X11 however, is another story
<spittnacid>
If i can get at least terminal with usb thru otg keyboard, that would be a great start
<spittnacid>
I can fumble my way around but im still learning. I dont expect some single click install and im guessing there may be quite a task ahead but could you just point me in a direction of where to start? Google is not my friend on this, all i get are shells on top of android results
<hipboi>
Turl: pong
<cheng>
try search in cnx-software
<spittnacid>
This may be over my head. Will i need to put u boot on it and in some way compile a flavor of linux with the neccasry drivers for my devices hardware? Not sure i understand.
hipboi has quit [Ping timeout: 264 seconds]
hipboi has joined #arm-netbook
hipboi has quit [Ping timeout: 240 seconds]
rellla has joined #arm-netbook
hipboi has joined #arm-netbook
<spittnacid>
I found information on the site provided but assume the available information is device specific. I understand that the onda vx610w has the same hardware as other devices. What is it comparable too that i can get specific information ?
<spittnacid>
Comparable to*
<spittnacid>
Thx for any help :-)
ssvb has quit [Ping timeout: 245 seconds]
<spittnacid>
A10-hwpack-bld.sh <device_name> is where i should be?
<rm>
I wonder if the inability to reboot is fixable in software
<rm>
(when a powered USB hub is connected)
ssvb has joined #arm-netbook
hipboi has quit [Read error: Connection reset by peer]
<mnemoc>
hipboi: i only wished that people from the community had the chance to catch some of the special offers before random people just find it via google
<lundman>
sweet
<hipboi>
ofcourse
<hipboi>
everybody here can go and support it
<mnemoc>
now? :)
<hipboi>
yes
<hipboi>
it's open anyway
<hipboi>
i can not stop it
<mnemoc>
cool
<mnemoc>
hipboi: please fix the delivery date.
<mnemoc>
october 2012 is kind of ... wrong
<mysteryname>
Supported :D
<mnemoc>
mysteryname: don't you have like a dozen already?
<lundman>
must. have. more.
<mysteryname>
There at Uni, and you have a sharp memory.
<mnemoc>
:)
<cheng>
can i have a link, to figure out what is discussing here
<mnemoc>
if you are booting from nand, you need to enable SUNXI_IGNORE_ATAG_MEM. if you are booting from uSD you need to have SUNXI_IGNORE_ATAG_MEM disabled
<Maqs>
k
<mnemoc>
SUNXI_IGNORE_ATAG_MEM was a very stupid choice of CONFIG_ name :<
<mnemoc>
it should have been CONFIG_SUNXI_ALLWINNER_BOOT_HACKS or somethings like that
<rm>
but now I'm reassured that ordering a couple of those slow overhyped <s>closed</s> devices was a good decision :p
<jinzo>
rm, depends on when you ordered them - I wish I would order them now. With all the HW improvements and my inability to do anything decent up to now (no time)
<jinzo>
but on the other hand, I'm itching to get a cubieboard
<ZaEarl_>
I'm having a hard time understanding what is open sourced. they're saying the stuff that "runs on the ARM"
<ZaEarl_>
which would mean the graphic driver isn't opened
<rm>
jinzo, just a couple of weeks ago, on the evening of the 512MB news
<jinzo>
rm, then you're lucky :D
<rm>
yeah
<rm>
I was not impressed with RS'es earlier "Despatch expected in 19 weeks", nor with Farnell's "1.5x of RS" price
<libv>
rm: it is full of BS.
<libv>
rm: the userland of the videocore is just a message and data passing layer
<libv>
rm: all the magic is running on the videocore
<jinzo>
still better than nothing.
<libv>
from basic modesetting to shader compilation
<rm>
> all of the VideoCore driver code which runs on the ARM
<libv>
jinzo: i could've freed the userspace in a matter of just a few weeks
<rm>
which is not much less than what you could really ask for
<libv>
jinzo: again, just message passing
<jinzo>
but yes, we'll see what actually comes out of it.
<libv>
rm: yes, but the statement is hugely overblown
<Gumboot>
It's all just message passing at some level anyway.
<libv>
the raspberry pi is one of the most closed platforms ever
<libv>
the videocore boots the arm core
<libv>
the videocore does all the modesetting
<rm>
libv, but now you can port the driver to new Xorg, old Xorg, Wayland, HaikuOS, etc etc
<libv>
the videocore compiles the shaders passed from userspace
<libv>
rm: it's a massive 2MB BIOS
<rm>
much better than when there was a binary driver for the latest Ubuntu Xorg only and that's it
<rm>
not really seeing how who you are would be all that relevant to the discussion :p
<rm>
what you are saying is correct
<rm>
all of it, I agree
<libv>
rm: i was one of the three guys who freed ati
<rm>
just highlighting that "it's still something", heck, "it's a lot"
<libv>
rm: i am also the one who is REing the mali
<libv>
i spent a few days looking at the videocore
<libv>
i kind of know what i am talking about here
<jinzo>
lima - oh, keep up the good work libv
<jinzo>
but noone is attacking your efforts and I believe everyone knows how weird the RPi is. Still, this is a step forward
<libv>
so ya'll stop your cheering, the raspberry pi only became slightly more open
<jinzo>
if not for technical reasons but for publicity/avareness.
<libv>
there is still 3 different versions of the binary blob
<libv>
each 2mb large, running some realtime os in the videocores own architecture
<Gumboot>
I thought the RTOS was public record.
<jinzo>
sounds like lots of fun.
<libv>
the arm really is there as a figlead
<Gumboot>
The ARM is there because that's easier than porting Linux to a new architecture.
<Gumboot>
And even if you did you'd still have the problems that MIPS face with application compatibility.
<libv>
freeing up userspace was a matter of:
<jinzo>
libv, if I can ask - what's your focus currently? Lima? something work related?
<libv>
a) very dark grey: disassembly, and just picking off the message ids and the way the data is formatted and putting this in a document online
<libv>
b) capturing message through LD_PRELOAD or from the kernel
<libv>
and it is just that, your essl shader is a few tokens and then the shader source sent to the kernel, which then puts it in a circular buffer that feeds it to the videocore
<libv>
jinzo: my current focus is some fun and games with my employer. hence me trawling through the disp code of the a10
<Gumboot>
libv: What's the difference between this and a highly competent hardware accelerator made of pure RTL?
<libv>
Gumboot: the hardware accelerator will have very limited scope if it is all in hw
<Gumboot>
Because it seems that all anybody has to do to squash this criticism is to put all the blobs on ROM so that nobody realises what's going on.
<libv>
Gumboot: it will be outdated pretty soon
<jinzo>
libv, happy to hear you're poking around. Hopefully the games will end in your favour :)
<libv>
the videocore is a dsp or a set of dsps
<libv>
with some code driving it, and that code is 2mb large, and will be full of bugs, and will not change with the times, even though the hw definitely is very flexible
<Gumboot>
Yeah, but why is that your business? There are other products on the market which may also be built on DSPs but which keep their firmware hidden so that nobody complains about them being undocumented. They just behave like a piece of hardware.
<libv>
Gumboot: i do not know, i have only been making it my business for the past 9ys
<libv>
i am a sick puppy and i enjoy doing what i do
<mnemoc>
:)
<Gumboot>
I have no problem with that. I'm not criticising your efforts (I can't possibly comment on your efforts). I'm just saying, nobody moans when the product keeps its firmware locked away.
<libv>
Gumboot: when the videocore doesn't want to drive your panel because you have a weird resolution that isn't supported by the message passing system, then do not come crying to me
<mnemoc>
if "firmware" you just push a bin and the world works. it's not the same with closed libs
<libv>
the world does not just work.
<libv>
firmwares are like bioses
<libv>
it's software that noone else is expected to look at
<libv>
all is well as long as all is well... but when something goes wrong, you're stuffed
<Gumboot>
mnemoc: Really, the bugs introduced by blobs in the Linux kernel aren't that far removed from the bugs introduced by blobs on hardware blocks, because a lot of that hardware is bus masters anyway.
<rm>
you have a weird resolution that isn't supported by the message passing system <- so like the A10 :D
<RaYmAn>
so, for the sake of completeless, considering the new rpi release, what makes rpi more closed than other SoC's?
<Gumboot>
They can make exactly the same mess, if not worse.
<libv>
rm: with the a10 it is shortsightedness in the kernel driver
<Gumboot>
I'm just saying... the only reason this product is open to criticism is because its firmware is visible.
<RaYmAn>
Mali400 presumably also has closed firmware bundled inside?
<libv>
RaYmAn: not that i know of
<jinzo>
RaYmAn, actually the whole infrastructure is a bit weird. The GPU that is running the closed FW get's up first and boots up the ARM core
<mnemoc>
jinzo: mali or r-pi's?
<libv>
Gumboot: fyi, i have also contributed extensively to coreboot and flashrom
<RaYmAn>
jinzo: yes. Incidentally, this is more or less what tegra2/tegra3 does as well.
<libv>
Gumboot: i am also the radeonhd developer who fought against atombios
<jinzo>
mnemoc, r-pi's, don't know that much about mali/a10
<RaYmAn>
libv: the mali400 doesn't have firmware?
<libv>
RaYmAn: not that i have noticed
* Gumboot
wonders at what point he's meant to either do full disclosure or back away.
<jinzo>
Gumboot, full disclosure is always good :)
<RaYmAn>
Well, it might be embedded/hardcoded, but I'd think there kind of has to be a firmware in there?
<libv>
RaYmAn: scheduling is done from userspace, and you have to build up the full commandstream from userspace as well
<Gumboot>
I used to work at Broadcom. I used to work at ARM. I now work at NVIDIA.
<lerc>
Your NDA beeper is making noise?
<Gumboot>
lerc: I think the collar is about to blow...
<RaYmAn>
libv: ok, so the difference is in what format you communicate with the GPU?
<jinzo>
Gumboot, oh goodie - nice career.
<jinzo>
Gumboot, on the other hand - we could get that information from linkedin
<Gumboot>
jinzo: Yeah, basically being a manifestation of the great evil you're all fighting. Sorry about that.
<libv>
RaYmAn: there is some hw to reset the different cores, but that is pretty much it
<jinzo>
so no worries :P
<jinzo>
Gumboot, I think we all realise it's not black and white like that.
<hno>
mnemoc, slapin, what?
<libv>
RaYmAn: writing the userspace driver is a lot tougher than for instance on the adreno
<libv>
RaYmAn: but it is not some insane convoluted mess like the pvr with the microcode, kernel, userspace all entwined
<hno>
ah, nand. Yes. but have not got it to work, most likely because I am not very familiar with NAND.
<jinzo>
Gumboot, your references/experiance are mind blowing.
<Gumboot>
lerc: It was actually you who described that movie to me, wasn't it? Wedlock, or something like that?
<libv>
what things will be like on the 6xx series, i do not know yet, but my view is that the mali200/400 is about as microcode free as it gets
<lerc>
Maybe. Was that a Rutger hauer one?
<Gumboot>
Yeah.
<RaYmAn>
I'm not too familiar with the inner workings of the SoCs/GPU's. So, basically what makes it more closed, in your oppion, is that it has a layer of closed-source code running that kernel communicates with? As opposed to direct access like on Mali400?
<libv>
RaYmAn: the adreno also runs a bunch of closed source ucode
<libv>
RaYmAn: the raspberry pi really is very very special in that it has a massive amount of "firmware"
<libv>
and that that firmware is responsible for 80% of what the machine does
<RaYmAn>
ok
<libv>
look at the code written for the a10 linux kernel, look at the insanity in there
<mnemoc>
:<
<RaYmAn>
It's kind of hard to tell which parts are allwinner being idiots and which part are ARM :)
<libv>
you will find similar insanity in every blob, especially bios or firmware, as there people do not expect others to see what is going on behind the scene
<cat_x301>
anyone to help here pls :) i wonder what might be wrong with this rule, mode is ignored for some reason..
<lerc>
I know most of my source is closed only for reasons of embarrasment.
<jinzo>
libv, also - people are under some preatty insane deadlines and that doesen't help either
<cat_x301>
mnemoc: this does not sound like an answer :D
<mnemoc>
cat_x301: nope, it's unrelated
<mnemoc>
cat_x301: free spam
<cat_x301>
:D
<libv>
jinzo: that is true for everyone, that is not the reason why i have seen so much crap from x86 bioses and atombios
<jinzo>
lerc, I tought of that too - then I open sourced a small Android app and when I got my first pull request. Man. That's dope.
<libv>
jinzo: atombios is the best example: ati sells the same hw in two different forms
<jinzo>
libv, it's true I mostly do web-developement but here there're firms that plan to release the libraries as open and count that in the time
<libv>
one consumer, one firegl
* cat_x301
went to ask on another channel..
<jinzo>
and therefore developers have a bit more time to write clean code, unit tests and so forth.
<libv>
the difference: firegl is slightly more conservative with power and clocking, a different set of fuses is blown for a different pciid
<libv>
and... then you go and compare the atombioses of consumer and firegl
<libv>
and see all the bugs they fixed
<slapin>
hno: mnemoc advertice you as A10's NAND expert
<slapin>
hno: so I have some questions re NAND workings
<libv>
and if some hw does not change for several generations, you can actually see even more evolution
<Gumboot>
lerc: Honeycomb was closed for that reason.
<libv>
rv630 atombios finally fixed a nice display blanking bug introduced in the first r500
<libv>
rv630 was like 3.5ys later
<lerc>
Gumboot: Yeah, That was probably an act of mercy.
<Gumboot>
They told some story about phones versus tablets, but they blew the secret with the ICS release notes.
<Gumboot>
There were many who were not granted that mercy. They had to suffer special privileges.
<slapin>
hno: I need to make u-boot and kernel MTD to work with A10's NAND controller, a problem I face now is how to directly control NAND chip with this controller, so put NAND command, put data, get response, on low level, so actually how to put data byte on NAND. This is needed for some operations required by MTD, and for me to understand how it really works.
* Gumboot
takes a moment to reflect on the horrors of software development.
<Gumboot>
Oh, I can save that for Memorial Day, I guess.
QingPei has joined #arm-netbook
<jinzo>
I would guess the hardware developement horrors triumph software ones :P
<libv>
at least when you limit yourself to the hw level, you only have to deal with the bottom black box full of errors
<Gumboot>
Those guys are a different breed.
<Gumboot>
Some of them wear ties.
<lerc>
That is one thing about limiting firmware. They have to get it right when they ship, no "fix it later"
<Gumboot>
Even when they don't have meetings.
<libv>
you do not have to deal with the bugs introduced with the layers above it
<libv>
lerc: this is not how it works
<lerc>
'course I have a laptop with a 945GM wich is just broken.
<libv>
lerc: they think they can always fix it in firmware
lkcl has joined #arm-netbook
<lerc>
libv: When do they give in and just ship FPGAs?
<Gumboot>
That's what I'd like to see being popularised instead of a generic ARM SoC.
<jinzo>
never probably. The performance/consumption ratio of FPGAs isn't that good.
<slapin>
lerc: when FPGAs in which reasonable hardware fits, will cost less than average car.
<Gumboot>
Instead of this whole "learn how computers work -- learn Python" thing, people could learn Verilog instead.
<jinzo>
Gumboot, we're just being tought FPGA/VHDL at my uni - on the Altera MAX+Plus II software (if I remembered correctly)
<jinzo>
Gumboot, I find FPGAs interesting, but learning it on a 10+years old software with old professors is killing all of my desire.
<Gumboot>
I'm sorry... do you really have a word in all capitals, the word "max", roman II, and _two_ plusses all in the same product name?
<Gumboot>
This is the sort of crap that's putting satirists out of work, you know.
<lerc>
championship edition.
<jinzo>
Gumboot, indeed :D
mysteryname has quit [Ping timeout: 252 seconds]
graffiti has joined #arm-netbook
<graffiti>
hi hipboi
<hipboi>
hi
<graffiti>
i am getting this error "Your payment can't be completed because one of the receivers can't accept payments at this time."
<graffiti>
when i try to make payemnt
<graffiti>
i have left comment in the comment section
<mnemoc>
Graffiti: yes, it's known. paypal found there was abnormal activity and blocked the account. hopefully will get solved soon
<graffiti>
mnemoc: Thanks will keep trying
<mnemoc>
Graffiti: because of time issues it's very unlikely it gets solved today
ZaEarl_ has quit [Ping timeout: 264 seconds]
<mnemoc>
Graffiti: already 20:30 in china
Hexxeh has quit [Remote host closed the connection]
jeremb has quit [Remote host closed the connection]
steev has quit [Remote host closed the connection]
Holo_ has quit [Remote host closed the connection]
Hexxeh has joined #arm-netbook
<graffiti>
mnemoc: guess so.. will check tomorrow :) hope "lucky dog" will be available.. still 9 remaining..
Holo_ has joined #arm-netbook
<mnemoc>
Graffiti: you can pay using credit card now. until paypal's BS is solved
<hno>
slapin, the A10 NAND driver is not related to MTD at all. But should be possible to make an MTD driver for it.
<hno>
slapin, the controller is command based and usually DMA driven.
lkcl has quit [Ping timeout: 246 seconds]
<slapin>
hno: I just need to check basic things, how could I run simple single command and read one or multiple byte reply? I don't need DMA stuff at this point.
<slapin>
RaYmAn: it is better than nothing. They found socially-accepted way to not disclose their source code, so you can upgrade kernels, use different distros, etc. I actually don't see any embedded 3D drivers open sourced anywhere. I think it is too much resources to develop such software.
<slapin>
mnemoc: any mplayer/ffmpeg patches?
<mnemoc>
slapin: not yet
<mnemoc>
slapin: but you are welcomed to start writting the bindings :)
<slapin>
mnemoc: I'd prefer jumping around NAND, I don't quite understand ffmpeg code
<mnemoc>
there is an xmbc tree with cedarx support, but i don't know any for other player/framework
<merbanan>
it is working for others also
<slapin>
xbmc makes device specialised media player which I don't want, and I also too conservative/lazy to change player...
<mnemoc>
what's the point in making a lib that only to wraps another?
<libv>
well, it wraps the other through the kernel and out to the videocore
<mnemoc>
ah, ok
<mnemoc>
libv: what is missing to "upstream" *cough* your libv_disp branch?
t0dbld1 has joined #arm-netbook
<mnemoc>
there are 3 competing forces working on the same driver currently :<
<libv>
everything, this is just the basis for starting real work
<libv>
hopefully later today i can add some infrastructure for register dumping and setting from userspace
<mnemoc>
libv: but those "minor fixes" can be merged already, right?
<mnemoc>
so I don't end up with 3 sets of conflicting patchsets :<
<libv>
i guess so, but they will make pulling new over-the-hedge dumps from allwinner impossible to merge
<libv>
there are similar cleanups for hdmi and lcd happening
<libv>
but i am a bit overwhelmed by the rpi noise
<mnemoc>
techn_: -----^
stefanro has joined #arm-netbook
cat_n9 has joined #arm-netbook
cat_n9 has quit [Client Quit]
pawel5870 has quit [Read error: Operation timed out]
QingPei has quit [Ping timeout: 268 seconds]
AndChat138129 has joined #arm-netbook
QingPei has joined #arm-netbook
QingPei has left #arm-netbook [#arm-netbook]
<gongon22>
can anyone help me
<gongon22>
i got
<gongon22>
a13B allwinner
<gongon22>
i flashed him with uberiod
<gongon22>
and got 1cm shifted TS on the right side
<gongon22>
how can i fix that
MindBeat has joined #arm-netbook
<gongon22>
can anyone help me ?
<slapin>
gongon22: calibration?
<techn_>
mnemoc: Yep.. I'll try to make that sun5i disp patch to ML.. And fix also identation/checkpatch problems :/
<techn_>
mnemoc: Yes.. I read Leith's patches
<mnemoc>
techn_: I made an stage/sunxi-3.0 starting with your patch, and then next_mali
<mnemoc>
techn_: will try to add libv's minor fixes on top of it
<techn_>
Whee.. cedarx libs :D
<mnemoc>
it was a busy morning
<techn_>
are they proven to work yet? :)
<techn_>
If they work we can merge next_mali! :)
<mnemoc>
it's in the stage branch already
<techn_>
nice :)
gongon22 has quit [Ping timeout: 244 seconds]
cheng has quit [Quit: Leaving]
bsdfox_ has quit [*.net *.split]
RaYmAn has quit [*.net *.split]
andoma has quit [*.net *.split]
Scepterr has quit [*.net *.split]
RaYmAn has joined #arm-netbook
andoma has joined #arm-netbook
Scepterr has joined #arm-netbook
bsdfox_ has joined #arm-netbook
bsdfox_ has quit [Changing host]
bsdfox_ has joined #arm-netbook
rellla has quit [Quit: rellla]
<techn_>
libv: nice work.. cat was also doing major disp refactoring :/
<cat_x301>
techn_: not really doing, i was going to do :)
<techn_>
cat_x301: good :D
<mnemoc>
techn_: but rebasing those upon stage/sunxi-3.0 is not been trivial :| .... would you?
<techn_>
mnemoc: which.. those ML patches or libv's?
<techn_>
or are they samething? :D
<mnemoc>
so we have 4 people working separatedly in refactoring the disp driver
* mnemoc
cries
<mnemoc>
techn_: libv's first. the 10x thing on the ML needs to be redone anyway
<mnemoc>
but PLEASE, start talking to each other :<
<techn_>
I have only EDID for HDMI left but that's still WIP.. and waiting disp to refactored.. and hopefully libv has time to check that it's done right
<techn_>
EDID==dynamic timings/resolution
<mnemoc>
as you know better the sun5i-import changes, can you rebase libv's upon stage/sunxi-3.0?
<mnemoc>
and push them
<mnemoc>
(into the stage/sunxi-3.0 branch)
<techn_>
hmm.. libv has refactore whole disp :D
<mnemoc>
yes
AndChat138129 has quit [Ping timeout: 240 seconds]
rellla has joined #arm-netbook
t0dbld1 has left #arm-netbook [#arm-netbook]
popolon has quit [Quit: Quitte]
<libv>
that's not refactoring, that's just beautifying
<libv>
i haven't even started sanitizing structure
<libv>
i still do not know how things fit together, but i do see lots of potential for restructuring already
<techn_>
libv: Rebased on sunxi-3.0.. next stage/sunxi-3.0
<libv>
relying fully on globals and passing "sel" is really ugly
<libv>
ok?
<techn_>
libv: to get rid of sell I thinked function pointers for different output types
<techn_>
*sel and sel-loops
gimli has joined #arm-netbook
<techn_>
or what you are thinking? :)
AndChat138129 has joined #arm-netbook
<mnemoc>
gimli: we got armhf/neon cedarx libs if you want to give them a try (in the case you haven't already)
<gimli>
mnemoc: have seen it. chaning my buildroot env to armhf is to much work
<mnemoc>
ok
<andoma>
mnemoc: i'll give it a spin later this week
<mnemoc>
andoma: thanks!
Quarx has quit []
AndChat138129 has quit [Ping timeout: 255 seconds]
<mnemoc>
my chromebook got shipped! \o/
<cat_x301>
mnemoc: happy happy mnemoc :)
<mnemoc>
armhf cedarx libs and chromebook the same day. happy happy joy joy
<cat_x301>
:D
<RaYmAn>
you got chromebook?
<mnemoc>
yes, the exynos5 one. shipping to a coworker in USA
<mnemoc>
and then forwarded as personal under 2kg package :p
<RaYmAn>
you actually got it in hand? :P
<RaYmAn>
(is what I meant to ask)
<mnemoc>
nah. amazon shipped it, 2m ago
<RaYmAn>
ah
<RaYmAn>
like that :)
<techn_>
I _need_ zaearl's gs811 ts drivers !
<techn_>
If he'll come here plz kick him
AndChat|138129 has joined #arm-netbook
lkcl has joined #arm-netbook
<hno>
slapin, you can read the data register with the cpu. No need foir dma.
<slapin>
hno: I want to know the sequence - wait for NAND ready, send command, write data, read data, some example would be nice. I can't still get to understand these 4 commands in registers, and random commands stuff...
AndChat|138129 has quit [Ping timeout: 255 seconds]
<slapin>
hno: I'd like to start from reading NAND ID as start, to establish a ground for later experiments
<hno>
slapin, then work from the existing function that does read the nand id.
<hno>
iirc it shows the sequence pretty well
<hno>
as you the only documentation I have is the sources.
<slapin>
hno: thanks! I will ask proper questions next time.
<hno>
hmm.. the function that reads nand id do not use the data register at all, instead it reads the result from nand controller sram.
<hno>
it's all blurry what I read in u-boot driver and what I read in kernel driver at the moment. They differ a bit in layout and I keep getting lost.
<slapin>
yeah, it is that ID, and it is read from SRAM, so wrong thing for me, because it prevents me from understanding how it works.
<slapin>
hno:
<hno>
slapin, not really. It's a good start regardless. The usage of _add_cmd_list is a key item to all I/O. And the sram is very important to the operation of this nand controller.
<hno>
slapin, can help digging a bit now. Both hands free again.
gzamboni has joined #arm-netbook
popolon has joined #arm-netbook
ChanServ has joined #arm-netbook
ol1ver has quit [Remote host closed the connection]
rellla has quit [Remote host closed the connection]
<RaYmAn>
It's likely it has been done to make it easier to implement userspace drivers, rather than just to prevent stuff from going public
<oliv3r>
so the VideoCore IV is in fact something like a gpu, but they don't wanna call it a GPU?
<oliv3r>
ohh, the drivers live entirly in userspace?
<oliv3r>
there's no kernelside drivers?
<oliv3r>
wouldn't that massivly hurt performancE?
<RaYmAn>
well, obviously it goes through kernel
<RaYmAn>
I'm probably bad at explaining
<oliv3r>
i'm even worse at comprehending, so it may not work :D
<mnemoc>
oliv3r: user space talks to kernel, kernel send a message to the video core requesting it to do the thing
<RaYmAn>
basically, what is usually implemented in userspace, is actually implemented on the videocore
<RaYmAn>
However, this release is big in some ways
<oliv3r>
aye, but normally, you have some kernel code and then libgl in userland
<mnemoc>
basically the "libgl" is implemented in firmware
<RaYmAn>
It does allow for anyone to compile GL drivers, X11 drivers, android drivers etc for whatever armel/armhf they so desire
<RaYmAn>
So that is important imo.
<oliv3r>
and together, that's the graphics stack. in the case of mali400; you have the opensource kernel side drivers, and the 'blob' driver
<RaYmAn>
But it's certainly an easy OSS release
<oliv3r>
i think fglrx is both blobbed;
<mnemoc>
it's scam
<RaYmAn>
it's almost the same on rpi, except -> userspace -> kernel -> videocore
<mnemoc>
they are intentionally misguiding people
<RaYmAn>
however, userspace is just a thin wrapper that passes the commands directly onto kernel which passes it to videocore, which then does the actual operations
<oliv3r>
so basically, it's a firmware 'shim'
<oliv3r>
e.g. glClear(); (from libGL) -> videocore which doesn't really do anything with it but passes it on -> kernel -> firmware, receives the glClear(); and executes it
<RaYmAn>
mnemoc: videocore is the part running the firmware, so almost right
<oliv3r>
well I guess, keeping the 'blob' in firmware, allows for a fully OSS system
<oliv3r>
i don't think having the firmware do that (much) is really good though
<RaYmAn>
indeed - it seems a lot like a hack to be able to say "fully opensource userspace"
<oliv3r>
what about kernelspace? isn't that now fully oSS too?
<RaYmAn>
sure
<RaYmAn>
but hasn't it been always?
<RaYmAn>
but it certainly is a lot easier to convince a company to open up this shim than it would be to actually open up drivers ;)
<mnemoc>
oliv3r: they made it "fully open source" without making anything interesting open
<mnemoc>
so, a marketing scam
<WarheadsSE>
yes and no
<WarheadsSE>
its semi-true, but remarkably obvious as semi-false to those actually in the know.
<mnemoc>
:)
<RaYmAn>
it may not entirely be a marketing scam though
<RaYmAn>
I mean, it is possible it was a valid technical design
<rellla>
mnemoc: the new cedar-libs - have there been any more files in the package?
<rellla>
i remember you mentioned here, that you've not uploaded a few?
<RaYmAn>
the marketing scam is mostly how they present it
<mnemoc>
rellla: some .o files
<mnemoc>
rellla: but they matched the .c files
<mnemoc>
RaYmAn: that's what makes an scam, the intention of fooling
ZaEarl_ has joined #arm-netbook
<mnemoc>
they indeed made it in a way userspace is free of closed libs
<mnemoc>
but they didn't opensource any GPU stuff
<RaYmAn>
which is commendable in itself
<mnemoc>
ZaEarl_: techn_ said he desperately needs the gt8xx something driver you offered
<techn_>
yes :)
<ZaEarl_>
great, I'll be posting it today
<techn_>
ZaEarl_: thanks :)
<techn_>
If I could make my tablet usefull
<oliv3r>
why would they design it in such a way
<oliv3r>
i know that some folks at some pie forum, claimed 'broadcom had some super duper interesting design into making their chip perform uber powerfull whilst being very power efficient'
<oliv3r>
and because of that, they would never open source anything
<mnemoc>
oliv3r: to protect their intellectual poverty^Wproperty and be able to counterattack TI
<oliv3r>
well 'ha ha', now we know their 'secret sauce' which isn't special at all
<oliv3r>
so then, A10 is still the most open source friendly SoC there is, and they don't even try
<slapin_nb>
hno, could you explain the role of sram with a10's nand controller?
<slapin_nb>
hno, I mean primitive operations, like reading a byte or two
<Sv>
why OABI?
<Turl>
Sv: I was just joking :)
<Sv>
oh
<Sv>
i still have an old jornada with debian sarge on a CF card, that thing was great back then
<hno>
slapin, the SRAM is used internally by the NAND controller to buffer data and ECC calculation. I think the data register is actually just reading/writing to SRAM and then the NAND controller does it's dance as needed on a per i/o block which seems to be 1KB.
<ZaEarl_>
techn_, gt811 driver posted to mailing list
<techn_>
ZaEarl_: thanks
<slapin_nb>
and how direct operations are done, like setting ase or cle and doing exactly one command with reply, how can I do it with direct IO? what is random read, is it mlc nand specific feature?
<slapin_nb>
hno,
<hno>
slapin, the NAND controller does the signal handshaking. You feed it a NAND command to execute.
<hno>
and the controller tells you when command finished, if you ask it to wait for finish.
<hno>
actually there may be up to 4 command codes in a single command execution. Haven't quite grokked how that works yet.
<hno>
0 to 4.
<slapin_nb>
and what is random read and write?
<slapin_nb>
I'm used to quite simple nand controllers
<hno>
slapin_nb, I think a reasonable example of single command execution is the read id command, or vender_get/set_param.
eFfeM has joined #arm-netbook
<hno>
slapin_nb, those show both how to set command and read/set data bytes, and without using DMA.
TheLarch has quit [Quit: gone]
<Marex>
hno: hey, MW is getting close to it's end ;-)
<Marex>
slapin_nb: hey
<Marex>
... did the world just shrunk again ?
<slapin_nb>
Marex, heheheh...
<slapin_nb>
Marex, hi
<slapin_nb>
what is MW?
<Marex>
merge window
<Marex>
for uboot 2013.01
<slapin_nb>
ah, it'd be nice to see some of allwinner stuff mainlined...
<mnemoc>
a guy from DENX added me on linkedin today... very weird considering i don't do anythin on u-boot-sunxi
<slapin_nb>
2013?
<Marex>
mnemoc: who ?
<Marex>
slapin_nb: yes
<mnemoc>
Marex: Stefan Roese
<Marex>
good guy tho ;-)
<slapin_nb>
it is only october
<Marex>
slapin_nb: yes, release every 3 months
<mnemoc>
Marex: know him?
eFfeM has quit [Quit: Leaving.]
<slapin_nb>
so even more ability to slack off than with kernel
<slapin_nb>
nice
<slapin_nb>
so there is some sunxi repo on git.denx.de?
<mnemoc>
not that I know of
<Marex>
mnemoc: in person, yes
gimli has quit [Quit: Verlassend]
<mnemoc>
Marex: :)
<Marex>
slapin_nb: we didn't even receive any patches, so what for
<Marex>
slapin_nb: what do you mean by more time to slack ?
<mnemoc>
Marex: so you are from DENX too?
<slapin_nb>
mnemoc, I know, just hoped for miracle
<mnemoc>
the spl code isn't mature yet. the cubie for example dies on reboot
<Marex>
mnemoc: I've been exceptionally lucky ...
<mnemoc>
Marex: :)
<slapin_nb>
Marex, three months is more than enough to slack, don't you think? no deadlines, etc.
<Marex>
slapin_nb: some patches arrive past the MW and are applied, but not often
<Marex>
slapin_nb: and no, we focus on making it stable, so no slacking
<slapin_nb>
Marex, ah, cool.
<mnemoc>
:)
<Marex>
slapin_nb: there's an internal CI server, which catches a lot, but not all
<Marex>
slapin_nb: then there's a testkit, but we can't run it on all possible boards
<mnemoc>
Marex: Stefan Roese mentioned he is receiving a cubie now :)
<Marex>
Stefan? I need to poke him tomorrow ;-)
<Marex>
interesting
<mnemoc>
:)
<slapin_nb>
btb, had dt thingy established? I have some samsung patches for some board, I need to light my burden
<mnemoc>
DT is mandatory these days
<slapin_nb>
I know
<slapin_nb>
but samsung is slacking with s3c24xx
<Marex>
rather it's a dead chip
<slapin_nb>
like half arm socs
<slapin_nb>
somebody needs to clear this mess
<mnemoc>
i thought simtec.co.uk had taken care of those
<libv>
is simtec still alive?
<libv>
some simteccers are now codethinkers
<mnemoc>
good question
<libv>
and another is now linaroing
<Marex>
I suspect it's about time for me to rebuild toolchains and go sleep while that runs
<mnemoc>
their website looks as 1998 as always
<Marex>
mnemoc: so what, my website is plaintext too and so what
<Marex>
better than funny flashing crap
<mnemoc>
Marex: you aren't a technology company :)
<slapin_nb>
damn I need somebody to help me make dt for samsung stuff
<Marex>
slapin_nb: you can contract me :p
<Marex>
slapin_nb: why the heck don't you ask in the LAKML ?
<Marex>
slapin_nb: but it's a long term work ... spaning a few release cycles
<Marex>
you're likely to hit 3.9 with small stuff, 3.10 with the rest
<Marex>
or so
<slapin_nb>
thats fine by me
<rzk>
<mnemoc> the spl code isn't mature yet. the cubie for example dies on reboot
<rzk>
same for mele a2000 v1.6
<Marex>
and I suspect the CPU is also pretty shitty because it's so cheap :()
<mnemoc>
o.o
<Marex>
mnemoc: what ?
<Marex>
mnemoc: just look at the docs :(
<mnemoc>
Marex: the docs are shitty, sure. because they expect manufacturers to pay them to design the PCBs and adapt the software
<mnemoc>
allwinner has been very clear that they are only interested in the chinese market
<mnemoc>
but that doesn't make the cpu shitty
<mnemoc>
their management is
slash_random has joined #arm-netbook
<Marex>
mnemoc: the CPU is pretty cheap, so I suspect not much went into proper debugging of the Ips
<mnemoc>
:)
<techn_>
mnemoc: disp rebase is also pretty hard .. allwinner has done alot same changes as libv
<mnemoc>
techn_: i have to admit i asked you after been trying for about 30m :|
<mnemoc>
but had to stop because i needed to resume $work$
<techn_>
I have done 2hours now.. and I think I did something wrong.. and have to start again
<mnemoc>
ouch
<techn_>
I'll continue tomorrow.. good night
<libv>
techn_: night
<libv>
mnemoc: who is running tablets with linux-sunxi today btw?
slash_random has quit [Ping timeout: 240 seconds]
<libv>
or are we all pretty much poking at the mele and other non-tablet devices?
slash_random has joined #arm-netbook
<mnemoc>
libv: at least Turl, ZaEarl_ and Quarkx are mostly android/tablet centric
<libv>
aha, ly-f1 is the best bet it seems
tzafrir_laptop has quit [Ping timeout: 276 seconds]
<mnemoc>
ly-f1 is the oldest A10 tablet
<mnemoc>
pre 1GB, pre IPS
<libv>
is anyone of turl, ZaEarl_, quarkx using an ainol device?
<mnemoc>
quarkx uses an aurora iirc
<ZaEarl_>
we do our own devices
slash_random has quit [Ping timeout: 240 seconds]
<libv>
i have an aurora2 which is quite nice, the aurora would be pretty tempting for me
slash_random has joined #arm-netbook
<libv>
it would actually be a shame to crack it open to expose serial though :)
<libv>
i know, sd card
<Turl>
beware with the ainol devices, because some of them aren't A10
<Turl>
:p
<libv>
i know, ainol are a bit slutty when it comes to sticking with one soc :)
<Turl>
there's even a MIPS one :P
<libv>
the aurora2 is destined for my 60y old mother who loves my main mali testing tablet for playing certain housewives games :)
<Turl>
haha :P
<libv>
i am too cocky about my knowledge on the mali to try my remali code on its dual fragment shader mali
<Marex>
libv: you're mucking with the lima driver ?
<Marex>
(it's remali now ? )
<libv>
but since i have a set of 7" haipads (telechips), a zenithink 7"er (amlogic), i can not help see the massive evolution these devices had in the past 2-3 years
<libv>
Marex: google for libv
<Turl>
libv: I have a msm7201a phone :P happened everywhere
<Marex>
libv: both "yes" and "no" are much shorter answer ;-)
<Turl>
libv: you might want to stick with gingerbread if you're going to use WIP mali libs :)
<libv>
Turl: my documentation is based on the telechips hw
<libv>
Turl: whenever a haipad under 50 comes by on ebay.de, i tend to grab it :)
<libv>
i already killed the battery on one (bad contact and constant charge discharge cycling) and the touchscreen (stupidity) on another
<libv>
they are running 2.3 androids
<libv>
and a very specific version too :)
<Turl>
you can use a keyboard on the broken touchscreen one :P
<Turl>
or mice even
<libv>
yeah, but android and a mouse is a bit of a pain :)
<libv>
i haven't had the heart yet to do some cutting/slashing :)
<Turl>
on ICS/JB it's as good as on a PC :P
<libv>
really?
<libv>
when i was poking around for intel a few months ago i was running ics on an eeepc
<Turl>
yeah, as far as I recall you get a mouse pointer thingy even :P
<libv>
that's even true for 2.3
<Turl>
not sure if 2.3 had it
<libv>
using a cursor to "emulate" touch is a bit... counter intuitive :)
<Marex>
libv: heh ... I had ICS on acer w500 ;-)
<mnemoc>
scroll by dragging with a mouse is painful as hell
<mnemoc>
at least arrows work... most of the time
<Turl>
libv: ics and higher requires samplerExternalOES / GL_TEXTURE_EXTERNAL_OES btw
<Marex>
libv: the presentation is good
<libv>
anyone know anything about ly-f2s tablet, is that basically a newer revision?
<libv>
of the ly-f1 that is
<libv>
i can get that one straight from .de, so i do not have to waste an afternoon to collect it at the customs office (like i had to for my mele)
<Marex>
libv: it's the same in .de, huh
<libv>
yeah...
<libv>
you declare the goods one place, then get given a pocket-change amount, then walk to another counter, where a man sits at a ledger and a small till, and then you watch your life tick away while he painstakingly writes the amount in the ledger...
tuliom has joined #arm-netbook
<libv>
looks like i will spend 80 on the ly-f2s, but on saturday, so my gf does not catch me receiving it and wags her finger at me for dragging even more crap into my room
popolon has quit [Quit: Quitte]
<slapin_nb>
libv: i run 3 tablets with A10 btw
<libv>
slapin_nb: which ones, what are you running on them, how do you get serial?
<Marex>
slapin_nb: you don't have a girlfriend !
<Marex>
libv: knew that problem
slash_random has quit [Ping timeout: 240 seconds]
<slapin_nb>
I run hyundai A7s with android kernel
<slapin_nb>
serial is obvious there
slash_random has joined #arm-netbook
<libv>
ok, so not the linux-sunxi kernel
<slapin_nb>
Marex, wanna talk about gfs? I do have two, and a member of american red cross llc in Russia
<slapin_nb>
libv: I run it on one of them, but have trouble with screen
<slapin_nb>
so I can't use it as tablet
<libv>
describe trouble, and the touchscreen or the lcd?
<slapin_nb>
hdmi works, lcd doesn't
<Marex>
slapin_nb: I don't want to talk about women ...
<Marex>
slapin_nb: I thought you were married
popolon has joined #arm-netbook
<slapin_nb>
Marex, I had that stuff in my head and was looking for occupation and my ex-gf recommended me to enter some volunteer-based org
<slapin_nb>
and I ended up doing nursing stuff
<mnemoc>
free software is volunteer-based too...
<slapin_nb>
I finally refused relocation to south Africa
<slapin_nb>
when you want to get rid of mental problems you are better for hard labor
<slapin_nb>
Marex, and don't worry, I'm still married
slash_random has quit [Ping timeout: 240 seconds]
<Marex>
slapin_nb: I see ... still marrier with two girlfriends ...
slash_random has joined #arm-netbook
<Marex>
that's so ... you
<mnemoc>
:)
<libv>
slapin_nb: so this is also a hyundai that's broken, how do you get serial, sdcard?
<slapin_nb>
serial is soldered, made small hole in case and got it outside
<slapin_nb>
not via sdcard
<Marex>
slapin_nb: I finally figured out who makes up for me in this part of life ;-)
<slapin_nb>
Marex, I'm very happy for you!
merbzt has joined #arm-netbook
<Marex>
slapin_nb: let's not continue this discussion, and stop hogging my happiness ;-D
<slapin_nb>
btw I have 1 year 3 months kid
<Marex>
oh my God
<slapin_nb>
not wanting to sleep now
<libv>
slapin_nb: aha, so the serial port can clearly be spotted on the mainboard?
<Marex>
slapin_nb: you have it with your wife of your GF ? ... or did you just split yourself in two ?
<slapin_nb>
yeah
<slapin_nb>
wife of course
<slapin_nb>
well, my new gf is still can't walk properly, as I am dedicated nurse for her via this weird org
<slapin_nb>
when I started she was a vegetable, addition to sofa using adult pampers
<slapin_nb>
now she can care of herself at home
<slapin_nb>
half year progress
<Marex>
ugh
<slapin_nb>
mending people's lifes is as fun as debugging software
<Marex>
I now see what you mean by GF
<slapin_nb>
well. she had no choice but to become my gf, I finally had to get some reward
<slapin_nb>
at
<slapin_nb>
at beginning she was a bitch full of bullshit, all volunteers stopped working with her because of continuous complains on abuse
<slapin_nb>
I had to handle all this
<slapin_nb>
sorry for noise
<libv>
slapin_nb: hrm, seems like i should get a userspace memory dumper up, not sure whether mapping /dev/mem will work though
<Marex>
libv: devmem shall work
<libv>
slapin_nb: if we get that to work under android, once we know enough about the display engine, we should be able to fix your tablet
<libv>
a "simple" matter of comparing before and after :)
<slapin_nb>
libv: I can try to uncover it to show how serial is soldered
<libv>
slapin_nb: would be very useful for the linux-sunxi site
<Marex>
gn
<slapin_nb>
gn
<slapin_nb>
libv: I can do it tomorrow, almoost 3am here, could have accuracy problems
<libv>
slapin_nb: no rush :)
<libv>
slapin_nb: i am going to buy the ly-f2s so i have some device with lcd for disp engine testing
<mnemoc>
:)
<libv>
slapin_nb: i hope that we will be able to fix the hyundai remotely :)
<slapin_nb>
I have a7hd with ips screen
<slapin_nb>
cool nice thing
<slapin_nb>
since hyundais sell out rooted, there might be chrooted linux distro with devmem installed...
<mnemoc>
all A10 devices come with `su`
<mnemoc>
that's a nice feature ;-)
<hno>
rzk, OK.
gsilvis has quit [Read error: Connection reset by peer]
slash_random has quit [Ping timeout: 240 seconds]
slash_random has joined #arm-netbook
<ZaEarl_>
ipad mini revealed. looks underwhelming to me.
slash_random has quit [Ping timeout: 252 seconds]
slash_random has joined #arm-netbook
<Turl>
ZaEarl_: it's a 'me too' device
<Turl>
apple is playing catch-up
<ZaEarl_>
poorly, too, which is good. :)
<Turl>
old processor, no GPS, low res screen, and they still managed to make it more expensive :P
<lundman>
ok
<lundman>
turl: so, wanna share your JB success? we should make a JB installer for mele