iamfrankenstein has quit [Ping timeout: 276 seconds]
<Tsvetan>
lkcl_ do not be ridiculous, all sources we have are made available to linux-sunxi developers, we just wait these to be merged in github and to make step by step instructions for the kernel compilation like we did for other boards
\\Mr_C\\ has joined #linux-sunxi
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
Soru__ has joined #linux-sunxi
Soru has quit [Ping timeout: 264 seconds]
panda84kde has joined #linux-sunxi
<oliv3r>
goooood morning
deasy has joined #linux-sunxi
<panda84kde>
mornin oliv3r :)
Quarx has joined #linux-sunxi
<panda84kde>
Hi guys. Does anybody know if any work has been done on the kernel to do correct initialization of mali resources for a20?
atiti has joined #linux-sunxi
wingrime has joined #linux-sunxi
<oliv3r>
lkcl_: the olimex images violating the GPL sounds really really odd. Olimex is very opensource focused!
<lkcl_>
Turl: they need to release the *exact* same source code. they've explicitly stated "we made some modifications and we are not going to release them"
<lkcl_>
that is an explicit GPL violation
<mnemoc>
they have the modifications in their github
<mnemoc>
but Tsvetan is probably the right person to answer that statement
<oliv3r>
they only put in some hacks to make the kernel buildable with their toolchains etc; anyhow, they want to push everybody to use the 3.4 A20 GPL kernel
<oliv3r>
the 3.3 kernel is a violating one to begin with
<oliv3r>
as the 3.3 one being mentioned there is with libnand if i'm not mistaken to start with
<mnemoc>
btw, is anything still missing in stage/sunxi-3.4 to merge that into sunxi-3.4 ?
<oliv3r>
ontop of that, tsvestan is basically saying 'don't split development resources in putting effort into the 3.3 GPL-violating crap kernel'
<mnemoc>
+1
<oliv3r>
so while technically it might be a violation (strong might as it violated before anyway), if its just some build fix hacks; i'm not caring too much. people who want to build the kernel can do the same hacks quite easily, otherwise, *beg* please move to 3.4
<oliv3r>
actually, i'll make a post on the olimex forum abou this
<oliv3r>
time to register ;)
hipboi_ has quit [Read error: Connection reset by peer]
<Tsvetan>
lkcl_ stop this crap please
<Tsvetan>
oliv3r we all know why lkcl_ is acting as ass
<Tsvetan>
we never hide anything from open source dcommunity - all info, sdks etc we got from allwinner immediately goes public
<Tsvetan>
right now Dimitar is downloading A20-SDK 2.0
<Tsvetan>
as the DVD they sent me was broken
hipboi has joined #linux-sunxi
<oliv3r>
hah, supprise
<oliv3r>
hipboi: hello!
<oliv3r>
hipboi: remember that I swaped my PHY on cb1, i think i saw it working for about 20 seconds, but it looks to be brokenagain, so I think something is wrong with the powersupply or something, no big deal, just annoying
rellla has joined #linux-sunxi
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
wingrime has quit [Ping timeout: 245 seconds]
wingrime has joined #linux-sunxi
Soru__ has quit [Ping timeout: 245 seconds]
<oliv3r>
Tsvetan: well i made a little post about why users should use the 3.4 kernel :)
<Tsvetan>
thanks oliv3r
_enrico_ has joined #linux-sunxi
<_enrico_>
hi
<wingrime>
Tsvetan: are you get new SDK?
<oliv3r>
Tsvetan: while i understand there are often less knowledgable users running all sorts of code grabbing stuff from random places, when getting a dev board to dev on, you shouldn't have any issue solving little problems
<Tsvetan>
wingrime - downloading with 10K/s 3GB image :)
<oliv3r>
hole shit
<Tsvetan>
hopefully will be complete today
<oliv3r>
i'd help, but dunno if i'd be helpfull
<Tsvetan>
china cloud server
<oliv3r>
your geologically closer to china then me :)
<oliv3r>
NL :)
<Tsvetan>
dimitar ran some client which handle broken downloads as it drops often
<oliv3r>
Tsvetan: how was your holliday?
<Tsvetan>
oliv3r shorter than I wanted to be
<oliv3r>
haha, i imagine
<oliv3r>
vacations always are
<Tsvetan>
BTW someone was talking how good is to work with well documented processor
<oliv3r>
Tsvetan: the A20? that's not that well documented :p
<Tsvetan>
AM3352 is very well documented ... and nobody wants to read 3000 pages user manual :)))
<oliv3r>
AM as in AMLogic?
<Tsvetan>
we try to run the second Ethernet port of AM3352 but its PITA
<oliv3r>
ohh TI AM3352
<Tsvetan>
and there are significant differences between AM3352 and AM3358 used in BBB
<oliv3r>
yeah, Ti is known to have really good documentation
<oliv3r>
what GPU does it have? powerVR?
<oliv3r>
SGX530 3D Graphics Engine
<oliv3r>
that's sad :(
<Tsvetan>
AM3352 have none IIRC
<Tsvetan>
its made for industrial apps
<oliv3r>
well that's for AM335x
<oliv3r>
in that case, it's coo,.
<Tsvetan>
the EVB we make have CAN and 2 Ethernet
<oliv3r>
nice server board :D
n01 has joined #linux-sunxi
<Tsvetan>
good for HMI
<oliv3r>
yeah :)
<Tsvetan>
and none of our other boards have working CAN nor 2 Ethernets so fills gap
<wingrime>
Tsvetan: we can connect ethernet over spi, and also we have CAN
<Tsvetan>
wingrime sure but in AM3352 you can do this without loading the processor
<Tsvetan>
where is the CAN driver for allwinner processors?
<wingrime>
Tsvetan: thats problem
<wingrime>
Tsvetan: but we have much strange stuff like ps/2 and pata
<Tsvetan>
Im not sure if Allwinner used the CAN driver at all, when I asked Wits for the Ethernet in A10S they said we made only tablets designs so far and never tested it :)) even had to ask Allwinner engineers it is inside the chip or just mistake in the datasheet
<wingrime>
Tsvetan: ps2 are regs are documented
<wingrime>
Tsvetan: there is stuff in chip they not want publish to be exsist
<wingrime>
Tsvetan: for example AC3 and DTS hardware decoders
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
<oliv3r>
Tsvetan: yeah AW soc's aren't as rosey as they appear to be
<oliv3r>
bugy untested 'secrative' stuff
<Tsvetan>
oliv3r they are very well pack with features for their price
<Tsvetan>
and for the last year Allwinner becomes more and more open
<Tsvetan>
when I compare now with Rockchip, we are at the beginning
<wingrime>
Tsvetan: I hope you can ask AW about some CAN docs
<wingrime>
Tsvetan: so We can do something, But I personaly have no idea where I can use CAN or debug it
<oliv3r>
wingrime: if we have hardware to connect with and someone who wants to do a driver, someone like tsvetsan surely can ask for docs
<oliv3r>
can is used a lot in cars
<oliv3r>
so a car PC that interfaces directly with its can bus is a possibility
<wingrime>
oliv3r: you can do it with GPIO
<Tsvetan>
lkcl_ the GPL warrior/protector should go and bug Rockchip who sell SDK based on GPL code instead to search for calf under the ox
<oliv3r>
Tsvetan: btw, on the next boards, solder the 32k crystal down (or hot glue it)! i keep forgetting to mention that ;)
<oliv3r>
Tsvetan: rockchip sells the SDK?!
<oliv3r>
wingrime: yeah but why bitbang it if you have hardware controller ;)
<wingrime>
oliv3r: there spi to CAN bridges
<Tsvetan>
oliv3r we found SMT 32kHz and move to them
<wingrime>
or i2c to CAN bridge
<Tsvetan>
sure, they asked me $5000 to give me the sources and datasheet
<Tsvetan>
Tom paid them 10 000 RMB for the same
<rah>
that's criminal
<Tsvetan>
:-)
<Tsvetan>
go sue them
<oliv3r>
Tsvetan: even better! i found it strange since you even had a solderpad under the can; i soldered it myself, just in case. I don't think anything really ever happens to the crystals though
<oliv3r>
10.000 RMB is how much USD or EUR?
<Tsvetan>
$1400 but this is the price for China
<Tsvetan>
if you are outside China the SDK cost between $5000 and $10 000 depend on how much money they decide can ask you :)
<panda84kde>
hi Tsvetan. Quick question: is the A20-SDK 2.0 you are referring to Linux or Android based?
<Tsvetan>
Android, Allwinner doesnt care for Linux
<oliv3r>
still a lot for usermanual + source
<oliv3r>
also means, that it's really hard for community to work on
<oliv3r>
but android is linux! :p
<Tsvetan>
oliv3r there are lot of people who do reverse engineering and already have picuntu - ubuntu running on RK devices
<oliv3r>
yeah but if ti's all REed, it slows development to a crawl
<Tsvetan>
patched linux to may run Android stuff
<Tsvetan>
I read somewhere that in future these Android patches may become part of the mainline but the process is painful as Android is weird Open Source project
<oliv3r>
half of it is mainlined allready
<oliv3r>
some things are still waiting to find there way
<Tsvetan>
Google do not share their intends nor allow other ppl to contribute freely
<oliv3r>
what I really want to see happen, is display framework unification, so we don't need android/linux video drivers that are so hugely different
<oliv3r>
Tsvetan: yeah android isn't that awesome if you think about it
<rah>
Tsvetan: have you reported them to gpl-violations.org?
<Tsvetan>
rah - nope, I have real work to do
<rah>
Tsvetan: that's a weak excuse
<Tsvetan>
rah now you know this info, what you will do about it?
<rah>
Tsvetan: I have no details, there's no much I can do
<rah>
s/no much/not much/
<Tsvetan>
rah I can provide you with these details
<rah>
Tsvetan: why provide them to me instead of to gpl-violations.org?
<Tsvetan>
rah I told you - I have real work to take care of
<rah>
that makes no sense whatsoever
<rah>
you offered to provide details
<Tsvetan>
yes
<rah>
providing details to gpl-violations.org will be no more costly in terms of time than providing them to me
<Tsvetan>
today I got cc ny lkcl_ who sent message how Olimex is violating GPL, I reply to this list but it say Im not registered
<Tsvetan>
I can forward you the e-mail where Rockchip ask me for $5000 to provide SDK sources if you have time and registration to this gpl list you can report them
<oliv3r>
Tsvetan: you know what the biggest problem is with people running 3.3 and 3.0.8 kernels?
<Tsvetan>
rah Im no lkcl_ and I do not want to waste my time I do remember same e-mails exchanges were one year ago for Allwinner, the result was few e-mail exachanges and nothing else
<oliv3r>
Tsvetan: they come to the ml and start asking all these questions or have all these problems, and expect help from us (or olimex/cubietech/everybdoy)
<rah>
Tsvetan: I don't understand
<rah>
Tsvetan: you offered to email them to me; why not email them to those addresses instead?
<rah>
09:29 < Tsvetan> I can forward you the e-mail where Rockchip ask me for $5000
<Tsvetan>
rah I told you been there done that one year ago, nothing happend
<Tsvetan>
if you have time and will to do this I can send you this e-mail, otherwise please do not bother
<oliv3r>
rah you don't report anybody to gpl-violations.org gpl-violations tries to help you get your code etc, they don't really do work for you
<rah>
Tsvetan: you sent an email from Rockchip asking for $5000 for source code to GPL software, to license-violation@gnu.org?
<panda84kde>
rah: It's pretty simple. Tsvetan doesn't want to be involved in any other step other than forwarding an email. If you want that email, then he'll forward you, and then you can send to license-violations, but all the subsequent replies and action are up to you
<oliv3r>
Tsvetan: i'm a member of the gpl-violatiosn list, and haven't seen it pass by, it's not passed moderation yet
<rah>
panda84kde: sending an email to license-violations doesn't obligate Tsvetan to be involved in any other step
<panda84kde>
rah: that seems like an "throw a rock and escape" approach
<rah>
panda84kde: no, it's more a "do something rather than do nothing" approach
<Tsvetan>
rah no its more than shit where you want to eat later ;)
<rah>
panda84kde: if Tsvetan doesn't have the time to engage with the process then he doesn't have the time, there's nothing anybody can do about that but he does have the time for forward one email, edidently, because he's offered to do it
<Tsvetan>
I work on Rockchip RK3188 board, and with or without Rockchip sources there will be Linux support for RK3188 soon
<rah>
Tsvetan: oh I see, you want to do business with these criminals gotcha
<Tsvetan>
I do not see reason to piss of Rockchip to make you happy :)
<rah>
Tsvetan: don't do it to make me happy, do it to make you happy
<Tsvetan>
rah if I knew there will be ever slight cosequence of this I would do it, I told you one year ago same conversations happend for Allwinner
<Tsvetan>
except the email exchanges NOTHING happend
<rah>
Tsvetan: between whom did the email exchanges take place?
<Tsvetan>
Allwinner keep releasing SDK with binary blobs, the only Linux support is done by Linux-Sunxi community
<rah>
Tsvetan: who was involved?
<Tsvetan>
I have to search e-mail one year ago and I got 300 emails per day please have mercy
<Tsvetan>
search the logs on the list for Allwinner
<rah>
Tsvetan: was there any party involved who has the power to enforce the GPL?
<rah>
like the FSF or gpl-violations.org or SFLC?
<Tsvetan>
I dont remember
popolon has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<rah>
Tsvetan: well, if you want to hide your identity, if you forward the email to me, I'll be happy to forward it on without any identifying information
<rah>
rah@settrans.net
<oliv3r>
now we're getting someone :)
<oliv3r>
good job rah ;)
<panda84kde>
rah: this channel is logged at the link in the title, so I think it's not a matter identity...
<rah>
o/ For great justice!
soul has joined #linux-sunxi
<oliv3r>
hno: tested the combined u-boot on "3" boards, cubie 512, cubie 1 and cubie 2
<oliv3r>
i don't have cubie 512 so used cubie 1 for it
<Tsvetan>
I think Rockchip do not hide that they sell their SDK, you can e-mail them at the web page and ask for it and get your quote :)
<_enrico_>
Tsvetan: do you have/can ask for additional documents (register guide etc..)?
<Tsvetan>
_enrico_ all info collected is on dl.linux-sunxi.org
<oliv3r>
_enrico_: The TV-Decoder? e.g. 'input'? I thought we had documentation for it
<oliv3r>
and drivers i belive we have too
<oliv3r>
_enrico_: i don't think there IS more then what we have :)
<_enrico_>
yes i was thinking that maybe there is someone to ask to
<_enrico_>
apart from a crappy driver and some registers there are no docs
<oliv3r>
it's not in the a20 user manual either?
deasy has joined #linux-sunxi
hipboi has quit [Remote host closed the connection]
<_enrico_>
only some base info (iomem, irq...), no detailed registers description
<oliv3r>
i'm not supprised
<oliv3r>
hno: can you see any reason why willis wangs' u-boot stops working? they all work for me here
<rellla>
hi, kernel config pro's here?
<rellla>
ok. i just ask: how can I create *_defconfig files without the default values - without manually editing it. so like sun4i_defconfig and not like sun7i_defconfig?
<mripard>
make savedefconfig ?
<mnemoc>
it will generate a "defconfig" file from your .config with only the not-default values which then you copy into arch/arm/configs/foo_defconfig
Black_Horseman has joined #linux-sunxi
<rellla>
savedefconfig \o/ thanks. one can't give it a better name ;) sun7i_defconfig seems to need some cleaning...
<rellla>
btw. stage/sunxi-3.4 doesn't build here
<mnemoc>
make sun7i_defconfig; make savedefconfig; cp defconfig arch/arm/configs/sun7i_defconfig
<mnemoc>
rellla: eh? now that I was going to merge it? :(
<oliv3r>
don't know if we want our devs poached :)
mouchon has joined #linux-sunxi
<oliv3r>
'and FREE FOOD'
<oliv3r>
also, you'd have to move there :p
<n01>
geee, 10+ 7+ 5+ ...
<mouchon>
hello i have simple question: I have an olimex A20 board can you point me to the banch to use for this board and a starting config file ?
<mouchon>
i don't full fonctional kernel, but be able to have multi UART, fine tune i2c bus frequency and fb for lcd
atiti has quit [Ping timeout: 245 seconds]
<mnemoc>
mouchon: stage/sunxi-3.4 sun7i_defconfig
<mouchon>
thanks a lot i will try it
<Turl>
"7+ years of firmware development experience in mass storage systems"
<Turl>
so basically you'd need to be an ex-WD employee or similar :P
<mnemoc>
so.... spam?
<Turl>
may be legit, but it's certainly not sunxi related
<mnemoc>
delete?
cajg has quit [Ping timeout: 245 seconds]
hglm has joined #linux-sunxi
<Turl>
do as you wish :P
<mnemoc>
:(
<Turl>
that reminds me I need to unsubscribe from arm-netbooks; it's been ages since I read anything relevant on it
<oliv3r>
mnemoc: delete it; don't need no poachers
<oliv3r>
is there anything related on it?
<mnemoc>
nope
<mnemoc>
deleted
<hglm>
It looks like one of yesterday's commits into stage/sunxi-3.4 breaks compilation for sun4i. sun7i does compile. After backing out the last 5 commits, it compiles and works fine on sun4i.
<oliv3r>
so regression finding time
<oliv3r>
hglm: does it fail to compile or fail to run'
<hglm>
It gives a compile error in the mali driver.
<hglm>
It seems to be related to the first commit of Tuesday (2b47db1501cfbe148dc7e722683eb0c910dca3d4).
lkcl_ has quit [Ping timeout: 245 seconds]
<mnemoc>
hno pushed those, his fault!
<boycottg00gle>
guilty
<oliv3r>
mali yuk
cajg has joined #linux-sunxi
<hglm>
BTW Has been any report of a A20-based tablet working (as opposed to dev boards)? I acquired one but didn't work on first try (have to be careful with new u-boot-sunxi though).
atiti has joined #linux-sunxi
<oliv3r>
hglm: a20 tablet, nice; 1G
<hglm>
Yeah it has 1G RAM, CPU at 912 Mhz, memory speed reported as 432 by script.bin.
<oliv3r>
geecko: with GPU disabled, that should give you some extra headroom
<geecko>
oliv3r, as long as it's stable, i guess it's good
<oliv3r>
geecko: but I wouldn't be supprised if 912 MHz isn't fully stable at 100% load :p only way to find out, is to do proper stress testing; the higher you clock, the better you test ;) but stable == stable
<oliv3r>
geecko: that link was for you too :) that's the temperature section
<geecko>
oliv3r, ok
<oliv3r>
'temp' is their wise name for temperature :)
<Turl>
oliv3r: on the touchscreen driver? really? :P
<oliv3r>
very confusing
<oliv3r>
but I haven't gotten it to work (I hadn't spent much time on it)
<geecko>
oliv3r, are you telling that 912MHz is not stable? dafuq?
<oliv3r>
Turl: really, there's 1 spot where it actually says 'temperature' in the manual
<oliv3r>
geecko: i wouldn't be supprised. how do you determin what is safe and stable?
<oliv3r>
geecko: you stress test it, then substract 10%
<geecko>
oliv3r, if it's default it's good
<oliv3r>
geecko: but what if your test is half-assed
<oliv3r>
or you run 100% but dont' stress each component
<geecko>
oliv3r, in fact it's running at 1056MHz since two weeks without a problem
<oliv3r>
or you leave out GPU out of your stress test
<geecko>
oliv3r, we won't use the gpu anyway
<oliv3r>
geecko: yeah it should have some headroom; i don't think it's that bad :) but in theory ... devils advocate here
<geecko>
oliv3r, 1100+ MHz wasn't stable though :)
<Turl>
bbl
<geecko>
gcc was crashing
<Turl>
geecko: yeah there's not much OC headroom
<geecko>
Turl, i think is was real 1156MHz
<Turl>
A10 starts crashing over 1.2GHz
<oliv3r>
geecko: same as a10 then ;)
<geecko>
the good news is : we are running our tests with 100% cpu for each core
<oliv3r>
good :D
<oliv3r>
so why are you porting mali for replicant?
<oliv3r>
just for fun? :)
<Turl>
other SoC vendors you can OC like 50% np (other than the heat)
<geecko>
but in real life, the processor won't be using that much
<geecko>
oliv3r, mali is a spare time project. i'm at work here :D
<Turl>
geecko: are you using ssvb's cpuburn? :P
<Turl>
I bet that leaves you with an unusable A20 (or crashes, best case scenario)
<geecko>
so we expect that the processor will be running at 408MHz much of the time
<oliv3r>
geecko: brickd runs on the a20 or on a remote?
<geecko>
oliv3r
<geecko>
the sbc with all the sensors plugged : idle 2W, burn 5W
<geecko>
still not crashed
<Turl>
oliv3r: if tmp is 0xff00 at start, and you just |= 0x4400 it'll still be 0xff00
<Turl>
if you mask and then or it'll be 0x4400
<oliv3r>
Turl: no i think they do tmp &= ~0xff00;
<oliv3r>
so they set those bits to all 00
<Turl>
oliv3r: 0xff00 &~ 0xff00 -> 0
<Turl>
| 0x4400 -> 0x4400
<Turl>
now, 0xff00 | 0x4400 -> 0xff00
<oliv3r>
wait let me re-read the code in question; but i'm pretty sure, that tmp &= ~0x1; is the same as 'clear bit 1'
<boycottg00gle>
whats the uboot installation method of the day?
<Turl>
oliv3r: indeed it is
<Turl>
oliv3r: so you need to clean potential cruft and then set whatever you like :)
e-ndy has joined #linux-sunxi
<oliv3r>
ah wait, i see my thinking mistake
<geecko>
wow, the memory modules are hot too
<boycottg00gle>
i guess it is: dd if=u-boot-sunxi-with-spl.bin of=/dev/sdX bs=1024 seek=8
<geecko>
oliv3r, i think it just failed
<geecko>
like, crashed :D
<oliv3r>
geecko: may be memory controller :)
<geecko>
oliv3r, running it again. 8 minutes so far
<oliv3r>
not hopefull :p
<e-ndy>
wingrime, ping hi
<geecko>
oliv3r, (i was playing with the heatsink and it _may_ have touched the PCB, hence the crash)
<geecko>
xD
<e-ndy>
wingrime, i see your name on sata for sunxi, is there any git repo for playing?
<wingrime>
e-ndy: nope still
<wingrime>
e-ndy: I need clock patches for it
<geecko>
12 minutes
<oliv3r>
geecko: haha, :p
<oliv3r>
geecko: upside is that it WAS hot allready, so you don't need to worry about warmup time too much
<oliv3r>
wingrime: where is your sata repo
<e-ndy>
wingrime, for which device? they are in 3.10 ( i guess a10)
<geecko>
my fingers >:(
<oliv3r>
geecko: feeling hot hot hot?
<geecko>
oliv3r, yes..
<geecko>
20 minutes and counting.
<oliv3r>
looks reasonable, lets see in 23-48 hours ;)
<geecko>
oliv3r, i don't think it's reasonnable :P
<geecko>
i can't put my fingers more than a second on it
<Turl>
geecko: now you know why it's called cpu burn ;)
<geecko>
Turl, lolz
<oliv3r>
geecko: waiting that long isn't reasonable? or the temperature
<geecko>
oliv3r, both xD
<geecko>
btw, thanks ssvb !
<ssvb>
geecko: you are welcome :)
<geecko>
ssvb, your blog seems really interesting too
<deasy>
hmm i always get weird caracters on ttl screen even when the cubie is already started
<geecko>
oliv3r, 30 minutes. ENOUGH!
<oliv3r>
geecko: not bad, but a proper burn in, 24-48 hours!
<geecko>
oliv3r, depends on the software too
<oliv3r>
cpuburn 24-48 hrs
<oliv3r>
if that runs, the freq, powersupply and memory timing is stable
<deasy>
i have see 41degrees on my cubie a10
<geecko>
with IntelBurnTest, 15 minutes is enough
<deasy>
(C)
<oliv3r>
41, that's cold :)
<oliv3r>
geecko: 15 minutes is never enough ;)
<geecko>
it's based on linpack
<deasy>
yes
<oliv3r>
geecko: 15 minutes is just enough to get the chip warmed up
<oliv3r>
geecko: but intel chips are manufactured to a much higher standard, and are rated much more appropiately
<geecko>
ok then :)
shineworld has quit [Remote host closed the connection]
<oliv3r>
hometime :)
<wingrime>
e-ndy: does not matter hw are smae
<wingrime>
*same
<e-ndy>
wingrime, ok
<e-ndy>
wingrime, is there way how can i contribute? :)
<e-ndy>
wingrime, to sata
mouchon has quit [Remote host closed the connection]
iamfrankenstein has quit [Ping timeout: 264 seconds]
<wingrime>
e-ndy: what actualy you want to do
mouchon has joined #linux-sunxi
<e-ndy>
wingrime, well i was thinking something like maintain patch queue for quilt/guilt/stgit in similar way like preempt_rt patch
<e-ndy>
wingrime, for all stuff needed to get it working even it not yet landed in mainline
<wingrime>
e-ndy: you can send patches to ML
<wingrime>
e-ndy: for linux-sunxi kernel
<wingrime>
e-ndy: also mainline will be nice as libata have not only on aw's socs
<mouchon>
hi, i have an issue with the stage/sunxi-3.4 on olimex a20 board with the A13-LCD7TS. The console is not displayed. The script.bin is the same that the one i used for the 3.3 kernel provided by olimex and with that kernel it work so can some one point me what can be wrong in my config ?
<wingrime>
mouchon: you don't forget set console= in cmdline?
<mouchon>
in the boot.cmd file ?
<wingrime>
maybe
<mouchon>
i will check thanks for the hint
<mouchon>
console=ttyS0,115200 this is normal if i understand, but howto activate terminal on LCD then
<wingrime>
mouchon: thats for uart
<mouchon>
here is my bootars: setenv bootargs console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait panic=10 ${extra}
<wingrime>
mouchon: you need more one for lcd
fredy has quit [Excess Flood]
<mouchon>
can you tell me which param i need to place ?
fredy has joined #linux-sunxi
geecko has quit [Quit: Quitte]
<e-ndy>
wingrime, well looking at old code, it should be doable in similar way like imx did it, instead of sw_ahci_platform module extend ahci_platform + platform-ahci-sunxi.c
_BJFreeman has joined #linux-sunxi
<e-ndy>
wingrime, but i'm c-lang noob :)
<wingrime>
e-ndy: it will be separated driver
_BJFreeman is now known as BJfreeman
<wingrime>
e-ndy: becose we have some PHY specific stuff
<wingrime>
e-ndy: we must not mix platform specific with otheres
<wingrime>
mouchon: may be something like console=/dev/fbcon but unsure
<mouchon>
ok will try but can i have both serial console and LCD terminal ?
<wingrime>
hno: ^
<wingrime>
mouchon: you can use console= twice
<mouchon>
ok
<wingrime>
mouchon: also don't foget fbcon must be enabled in kernel config
<mouchon>
ok i will check this
<wingrime>
mouchon: try also console=tty0 or console=tty1
<mouchon>
ok frame buffer console support is enabled and builtin, will try the different way. Thanks a lot
pacopad has joined #linux-sunxi
<pacopad>
Hi All , i compiled the g2d on linux 3.4.43 kernel for sun7i
<pacopad>
it loads , create /dev/g2d , Xorg feels happy to find it , according to logs
<pacopad>
how can i really try it ?
<e-ndy>
wingrime, as that phy is sunxi specific, it make sense to make phy specific stuff in arch/arm/mach-sunxi/devices/ like in case of mach-imx
<e-ndy>
wingrime, and device tree binding to load ahci_platform driver
<wingrime>
e-ndy: in other hand there much devices that have plaform specific part in other folder
<rellla>
oliv3r, mnemoc: It was mentioned on ml already, that building stage/sunxi-3.4 breaks with ARCH_SUN4I in mali module. I've sent a patch, but i've not tested, if loading mali module might crash.
<rellla>
so no gcc issue
stulluk has joined #linux-sunxi
<stulluk>
Hi, does anybody know, How to modify sys_config.fex file to enable RGB LCD instead of HDMI mode ?
<stulluk>
But when I compile with this settings, I can't get output on LCD, it only outputs at HDMI...
<stulluk>
How to modify it?
* boycottg00gle
would be happy to get hdmi working
<boycottg00gle>
stulluk: hdmi works for you?
<stulluk>
Hi, yes, my olimex A20 board works with HDMI by default
<stulluk>
Strange point is
<stulluk>
If you see the default sys_config.fex above,
<stulluk>
it is set to LCD, not HDMI
<stulluk>
I think I am missing some basic clue, but I am not sure why...
<boycottg00gle>
stulluk: maybe you use kernel parameter?
<boycottg00gle>
stulluk: cat /proc/cmdline
<stulluk>
What kernel parameter? Can you explain more clearly?
<boycottg00gle>
disp.screen0_output_type=3
<boycottg00gle>
maybe in uEnv.txt
<stulluk>
It is A20 ,
<stulluk>
ok wait
<boycottg00gle>
unfortunately have to leave now :(
<boycottg00gle>
stulluk: did you build your own image?
<boycottg00gle>
stulluk: which device is that (i have a olimex a20)
<stulluk>
I build my own Android image
<stulluk>
by using A20-SDK
<stulluk>
I fixed ethernet issue
<boycottg00gle>
ah android
<stulluk>
I added I2C tools package
<stulluk>
my proc/cmdline is: console=ttyS0,115200 root=/dev/system init=/init loglevel=4 partitions=bootloader@nanda:env@nandb:boot@nandc:system@nandd:data@nande:misc@nandf:recovery@nandg:cache@nandh:databk@nandi:UDISK@nandj
<stulluk>
So, there is nothing related with HDMI or LCD
<stulluk>
By the way, A20-SDK doesn't have any uEnv.txt ?
<stulluk>
I think that file is for linux builds, am I correct?
notmart has quit [Quit: notmart terminated!]
<boycottg00gle>
i don't know
<boycottg00gle>
i try to get hdmi working with linux build
boycottg00gle has quit [Remote host closed the connection]
n01 has quit [Ping timeout: 245 seconds]
ganbold_ has quit [Remote host closed the connection]
<Turl>
hm, some french guy is replying to all french posts ever on the ML saying something :/
ZetaNeta has quit [Read error: Operation timed out]
atiti has joined #linux-sunxi
<deasy>
Turl, interesting :p
<oliv3r>
e-ndy: wills wang did the sata driver pretty much allready
<deasy>
hmm ok he floods
<oliv3r>
mouchon: wingrime we don't have console on fb in u-boot; kernel should be possible
<mouchon>
oliv3r: don' need it for uboot but in kernel. In the olimex r2 image this work well , i try to have it with stage/sunxi-3.4. Should be a configuration missing somwhere but not able to find it. I try to add console=/dev/fb0 or 1 but this doesn't work
<mouchon>
other question there is a way to change the i2c busses frequency to 50khz ?
<mnemoc>
stage/sunxi-3.4 should be fixed now
<oliv3r>
Turl: do we have many french posts at all?
<oliv3r>
mnemoc: what ws the big issue?
ZetaNeta has joined #linux-sunxi
<oliv3r>
mouchon: i take it its done with the i2c-tools
<mnemoc>
oliv3r: missing include
<oliv3r>
that's not too bad
<oliv3r>
make sure to punish the offender badly!
<mnemoc>
nah
<mnemoc>
it's my fault to not take proper care of the ML :(
<wingrime>
mnemoc: also now we can move ml to our server
<mnemoc>
on the TODO but not on the top
<mouchon>
oliv3r don't see any clock related param in i2c-tools
<oliv3r>
mouchon: how do you do it with any other driver?
<oliv3r>
mnemoc: punish yourself!
* mnemoc
takes the leather suit and whip himself
<mouchon>
never had the chance to doit with oher driver. I am noob with this part of linux
akaizen has joined #linux-sunxi
<mouchon>
this is first time i played with this kind of board, i mainly used ATMEL SAM3S based boar without embeded linux.
<oliv3r>
ah, well i've never used i2c-tools
<oliv3r>
mouchon: so i don't know how i2c is done from a users point of view, but I cannot imagine that a user cannot change the freq of the bus, since there's various speeds, 125, 400k etc
<oliv3r>
oth, it's quite possible there's only those 2 speeds supported, and a 50khz freq. could be 'custom' and would need a hacked driver
<oliv3r>
mripard: did the i2c driver for the mailine kernel; so maybe he knows :)
<mouchon>
i looked into kernel driver and it seem that possible value are 100khz and 400khz
<mouchon>
but don't know this freq is fixed
<mouchon>
howto
<Turl>
I think there's an option on fex for it
<mouchon>
i will look at this
<oliv3r>
mouchon: why do you need 50Khz anyway? Isn't it non standard? I thought only lowspeed (125kHz) and hi-speed (400 kHz) are the official speeds
<oliv3r>
mouchon: as far as your uart question goes, i think you have quite a few to choose from :)
<mouchon>
because i have a non standar chip that use this frequency
<oliv3r>
1 uart is routed to the uart pins, 1 uart is shared with the uSD breakout
<oliv3r>
mouchon: if your lucky, you can adapt it easily using fex, if unlucky, you have to hack the driver to add that supported freq.
<oliv3r>
mouchon: if all you need is simple TX/RX uarts, there's one shared with SPI1's MOSI/MISO on PortA
<oliv3r>
hmm, that one might not be routed out, as it shares pins with the emac
<Turl>
hm nope, nothing in fex
<Turl>
from what I can see it's hardcoded in the plat data
<oliv3r>
rellla will be perfectly suited for testing ;)
<wingrime>
jemk: make it somehow usable like I described
<wingrime>
jemk: thats indeed ugly
<wingrime>
jemk: you get from vdpau XDrawable ?
<jemk>
wingrime: Display and XDrawable
<wingrime>
ssvb: how get window size from XDrawable ^^
<ssvb>
wingrime: something like "man XGetGeometry"
<rz2k>
if anyone is going to test Mali-400 blobs from Benn @ mailing list - only the non-monolithic one is going to work, other one will give you eglgetdisplay() errors.
<wingrime>
ssvb: witch color can be used for overlay color key?
<wingrime>
for avoid windows overlap
<ssvb>
wingrime: you can configure any color
<wingrime>
ssvb: it must be unused
<ssvb>
well, it's one of the problems with colorkey
<jemk>
wingrime: there are many more things missing for nice output, i shortcut the whole way vdpau handles output
<ssvb>
with 16,777,216 color variants, the probability of a collision is reasonably low
<wingrime>
ssvb: also question , if I request XV from you driver and will not send any data , overlay will be configured automaticly? general idea not send any data to driver and simply use correct configured overlay for output
<jemk>
wingrime: this is still only for decoding testing and not a real player for daily use
<wingrime>
jemk: yes, I know
<jelly-home>
ssvb: unless you do like nvidia did and pick (0,0,0) for the key
<ssvb>
jelly-home: :)
<wingrime>
jemk: but we can't go next untill we get bug reports, also, for dealy use , all configuration stuff must be packed to structures and send to kernel side driver , but lib will be still in place
<wingrime>
jemk: process will be same , but we add more middleware
<wingrime>
jemk: mainline in way, but this PoC are will be only linux-sunxi kernel
<wingrime>
oliv3r: How about write manual and anonce
n01 has quit [Ping timeout: 264 seconds]
<wingrime>
oliv3r: phoronix
<oliv3r>
oh yeah
<oliv3r>
well let rellla try to use it
<oliv3r>
with help frmo you guys
<jemk>
wingrime: did you even test if it works before these ideas? i only testet on one system(a10)
<oliv3r>
he could write the basic manual (since he's not a coder and doesn't know 'what to do') so he's fairly unbiased
<oliv3r>
once it works; lets get in touch with phoronix
<oliv3r>
well still worth an anouncement, if you need testers :)
<wingrime>
jemk: it does not matter, for first version
<oliv3r>
it's more like 'look world, here's what is possible today!'
<wingrime>
oliv3r: thats anyway cool news
<wingrime>
lkcl_: ping
<wingrime>
jemk: also, do you remember a13 reg that I had ? thats reg do startcode find
<wingrime>
*search
<wingrime>
lkcl_: ping
<oliv3r>
wingrime: lkcl_ only checks here very very rearly
<jemk>
wingrime: yes, at least for h264 it also exists on a10, but i found it easier to do this in software for now, would only introduce extra error sources
<wingrime>
jemk: If I add qt here and code, it will be look like wireshark
<wingrime>
mnemoc: ping
<jemk>
wingrime: maybe you could find out how to check for finish in the get_ue functions to get rid of this usleep. blob queries status reg, but i don't know which bit
Black_Horseman has joined #linux-sunxi
<wingrime>
jemk: give me time
notmart has quit [Ping timeout: 264 seconds]
tinti has quit [Quit: Leaving]
tinti has joined #linux-sunxi
tinti has quit [Changing host]
tinti has joined #linux-sunxi
<wingrime>
jemk: after any trigger start , ve send irq to kernel driver, if you wait in VE_WAIT irq you get event and return
<wingrime>
jemk: about flags
<jemk>
wingrime: no, this basic bit reading doesn't trigger interrupt
<wingrime>
jemk: I get flags for MPEG one , but h264 still not
<lkcl_>
wingrime: ack
<jemk>
wingrime: and waiting for interrupt would be overkill for such short jobs
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<lkcl_>
have you told dr stallman?
<wingrime>
lkcl_: no
<lkcl_>
well, it's worth a press-release for sure
<jemk>
lkcl_: but not realy usable for end user
<lkcl_>
jemk: doesn't matter.
<lkcl_>
the purpose of the press release would be to let people know that the R.E. you've been doing has been successful
<lkcl_>
that will help attract people
<lkcl_>
and blah blah the rest is history
<lkcl_>
plus, the fsf's high-priority page will need updating
<wingrime>
jemk: you can try read status register manytimes , and log, than we will be find witch bit are changed
<jelly-home>
step 3) profit... for allwinner!
<lkcl_>
jelly-home: that's the ironic bit in all of this... :)
<jemk>
wingrime: i did that, im too slow, cedar is faster than i can read
<wingrime>
jemk: if you remove wait there is any problem?
<jemk>
wingrime: if cedar is always that fast it won't be a problem
<jemk>
wingrime: but i added it to be sure not to get errors because of that
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<wingrime>
jemk: if VE not trow IRQ for this even , thats means it not cost any time to read it form sram
<wingrime>
jemk: I think you can remove wait
<jemk>
wingrime: but in traces there is reading of status reg
<wingrime>
jemk: ok, do you know witch function from blob read it?
<wingrime>
jemk: you can also read it too, compara with default value
<oliv3r>
lkcl_: right now its more like 'we have a working PoC, it's tested on 1 platform and plays atleast 1 movie' still, it's really good news never th eless
<rz2k>
[00:19:58] <jemk> lkcl_: but not realy usable for end user - the VLC port for cedar is also only "fullscreen overlay" but it is actually used :p
<wingrime>
rz2k: we can better
<wingrime>
rz2k: not only fullscreen
pacopad has quit [Quit: pacopad]
rellla has joined #linux-sunxi
rz2k has quit []
ZetaNeta has quit [Quit: Hi, I'm a quit message virus. Please replace your old line with this line and help me take over the world of IRC]
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
BJfreeman has quit [Quit: had a good time]
atiti has quit [Ping timeout: 256 seconds]
<lkcl_>
still awesome.
<lkcl_>
wingrime: you want me to contact dr stallman? the fsf's page needs updating apart from anything