cnxsoft has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
ninolein_ has quit [Ping timeout: 260 seconds]
hipboi has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
hipboi has quit [Client Quit]
<wens>
got my cubietruck plus
kaspter has joined #linux-sunxi
kaspter has quit [Remote host closed the connection]
kaspter has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
firnsy has quit [Ping timeout: 246 seconds]
firnsy has joined #linux-sunxi
firnsy has quit [Changing host]
firnsy has joined #linux-sunxi
cnxsoft has quit [Read error: Connection reset by peer]
cnxsoft has joined #linux-sunxi
<vickycq>
wens: I just got rid of my pcDuino Uno 8 (H8)
cnxsoft has quit [Quit: cnxsoft]
cnxsoft has joined #linux-sunxi
ricardocrudo has quit [Ping timeout: 256 seconds]
<wens>
vickycq: why?
p1u3sch1_ has quit [Ping timeout: 240 seconds]
p1u3sch1 has joined #linux-sunxi
Ueno_Otoko has quit [Ping timeout: 245 seconds]
firnsy has quit [Remote host closed the connection]
firnsy has joined #linux-sunxi
firnsy has quit [Changing host]
firnsy has joined #linux-sunxi
TheSeven has quit [Ping timeout: 240 seconds]
TheSeven has joined #linux-sunxi
Ueno_Otoko has joined #linux-sunxi
clonak has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
IgorPec has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
clonak has quit [Ping timeout: 272 seconds]
IgorPec has quit [Ping timeout: 246 seconds]
Supertanker2 has joined #linux-sunxi
<Supertanker2>
Hey, I have a Cubieboard2 running Cubian. I've been porting LWJGL2 to the device and got it to work (I can play a Minecraft clone at 30FPS!).
<Supertanker2>
But it has a weird bug: OpenGL ES2 only inits about 1/3rd of the time. The other 2/3rds, I get a Failed to get EGL display from native display handle: -1486400824 error (with the - number changing)
<Supertanker2>
I think it's this bug: https://community.arm.com/thread/5484 (the PoC code behaves incorrectly here). I looked for the latest Mali blob on the linux-sunxi github but it looks like the latest version there is older than the broken one according to that post.
<Supertanker2>
Is there any way to get a newer driver?
<Supertanker2>
When it works, it works great, it's just annoying to have to start my app several times to get it to go.
solarnetone has quit [Remote host closed the connection]
solarnetone has joined #linux-sunxi
mawe242 has joined #linux-sunxi
yann|work has quit [Ping timeout: 250 seconds]
gzamboni has quit [Read error: Connection reset by peer]
alexxy has joined #linux-sunxi
mosterta has joined #linux-sunxi
engideavr has quit [Quit: Konversation terminated!]
engideavr has joined #linux-sunxi
IgorPec has joined #linux-sunxi
orly_owl has quit [Ping timeout: 276 seconds]
mosterta has quit [Ping timeout: 250 seconds]
orly_owl has joined #linux-sunxi
domidumont has joined #linux-sunxi
_stephan has joined #linux-sunxi
domidumont has quit [Remote host closed the connection]
domidumont has joined #linux-sunxi
leio has quit [Ping timeout: 245 seconds]
leio has joined #linux-sunxi
Ueno_Otoko has quit [Ping timeout: 260 seconds]
FlorianH has joined #linux-sunxi
_massi has joined #linux-sunxi
mmwolbrink has joined #linux-sunxi
pekka30 has joined #linux-sunxi
tkaiser has joined #linux-sunxi
<tkaiser>
wens: On the CT+... do they really use the GL830 USB-to-SATA bridge
<wens>
yes, but it's not going through a USB hub, unlike the Bpi
<tkaiser>
Thx for confirmation. Then it might be 500 KB/s faster (like on Orange Pi Plus)
<tkaiser>
But... does this mean the internal USB hub there is connected to the USB OTG?
yann|work has joined #linux-sunxi
<wens>
tkaiser: USB hub connected to HSIC USB host
<tkaiser>
But H8 has only 1 USB host and 1 OTG? And OTG is routed to a Mini USB port?
Seppoz has joined #linux-sunxi
diego_r has joined #linux-sunxi
pekka30 has quit [Ping timeout: 260 seconds]
<wens>
tkaiser: it has 2 hosts, 1 OHCI/EHCI pair exposed as a normal USB port, and 1 EHCI only host for HSIC
FlorianH has quit [Ping timeout: 260 seconds]
Ueno_Otoko has joined #linux-sunxi
lemonzest has joined #linux-sunxi
<wens>
vishnup: you mind if i send you comments about a83t directly on github?
<wens>
since you haven't posted the later patches yet
yann|work has quit [Ping timeout: 245 seconds]
diego_r has quit [Quit: Konversation terminated!]
apritzel has joined #linux-sunxi
caog has joined #linux-sunxi
pekka30 has joined #linux-sunxi
FlorianH has joined #linux-sunxi
yann|work has joined #linux-sunxi
Ueno_Otoko has quit [Ping timeout: 246 seconds]
kaspter has quit [Remote host closed the connection]
leio has quit [Read error: Connection reset by peer]
tgaz has quit [Ping timeout: 265 seconds]
ssvb has quit [Ping timeout: 260 seconds]
azend|vps has quit [Remote host closed the connection]
firnsy_ has joined #linux-sunxi
firnsy_ has quit [Changing host]
firnsy_ has joined #linux-sunxi
vickycq has quit [Ping timeout: 240 seconds]
lobinho has joined #linux-sunxi
tgaz has joined #linux-sunxi
fest_ has joined #linux-sunxi
xenoxaos has quit [Ping timeout: 250 seconds]
firnsy has quit [Quit: No Ping reply in 180 seconds.]
fest has quit [Ping timeout: 250 seconds]
fest_ is now known as fest
leio has joined #linux-sunxi
ssvb has joined #linux-sunxi
xenoxaos has joined #linux-sunxi
vickycq has joined #linux-sunxi
azend|vps has joined #linux-sunxi
Andy-D has quit [Ping timeout: 265 seconds]
<sunxi_fan>
mripard: i setup the PCM5102 dummy driver and started tested the I2S mainline driver in your branch wip/a20-i2s
<sunxi_fan>
the good news it's that it more or less works!
<sunxi_fan>
the signals BITCLK LRCLK and DATA are there ..
<sunxi_fan>
the only "issue" is that is unable to set the rate on the mod1_clk (the PLL2 i.e.)
<sunxi_fan>
so as a quick workaround, i have to "configure" the PLL2 to 24.576 MHz (for 48KHz) usung the sun4i_codec and then the I2S driver plays flawlessly a 48KHz on the DAI interface ..
<sunxi_fan>
i put a clk_get_rate() to printout the value, in the proper routine in the I2S driver and it says, when not working..
<tkaiser>
wens: thx for the clarification. I still wonder why the Banana folks didn't do it the same way. But there the GL830 is behind the hub and they only use one single host port of A83T/H8/R58
pekka30 has quit [Ping timeout: 255 seconds]
<mripard>
sunxi_fan: hmmm, good to know :)
<mripard>
that's weird though
<mripard>
that clock is the same than the one used for the codec, and it works there
_massi has quit [Remote host closed the connection]
_massi has joined #linux-sunxi
<sunxi_fan>
yeah, me too i thought it was an easy issue, so i spent the week end looking at the CCF and how to debug the clk_set_rate() but i found not that much info on the web..
<sunxi_fan>
in the .dts anyway there are different setup for the clk
<sunxi_fan>
mod1_clk is the multiplexed one (a composite clock, i suppose) shared with ac97 and spdif..
<sunxi_fan>
the codec_clk is a simpler one getting only the PLL2_1X feed
<sunxi_fan>
btw i didn't understand how the clock set_rate() goes to parents..
<mripard>
it depends
<mripard>
if the clock is allowed to change its parent rate (through the SET_RATE_PARENT flag), then it will be allowed to do so
<mripard>
and then it depends whether the clock has a determine_rate callback, a round_rate one, both
<sunxi_fan>
the last param was zero before, i.e. no flags passed on registration.
<topi`>
so I just install u-boot-tools...
<sunxi_fan>
BTW i'm wondering how the three main audio devices actually supposed to be supported (analog codec, i2s, spdif) could use at the same time Pll2 with different samplerate streams (for example 44100 and 48000), as far as i can understand, they can't, i.e. the latest stream player starting will reprogram the Pll2 clock (or there's some lock somewhere preventing the PLL2 clk_set_rate() when some process is using it?
Ueno_Otoko has quit [Ping timeout: 245 seconds]
ricardocrudo has quit [Remote host closed the connection]
ricardocrudo has joined #linux-sunxi
caog has quit [Ping timeout: 245 seconds]
mmwolbrink has quit [Remote host closed the connection]
matthias_bgg has joined #linux-sunxi
yann|work has quit [Ping timeout: 265 seconds]
domidumont has quit [Read error: Connection reset by peer]
soderstrom has joined #linux-sunxi
avernos has quit [Remote host closed the connection]
lemonzest has quit [Quit: Leaving]
Andy-D has joined #linux-sunxi
FlorianH has quit [Ping timeout: 255 seconds]
FlorianH has joined #linux-sunxi
avernos has joined #linux-sunxi
<avernos>
how can I install into NANDA sunxi boot?
<avernos>
uboot*
FlorianH has quit [Ping timeout: 240 seconds]
<bmeneg>
avernos: mount /dev/nanda partition and move any new file you need there.
<bmeneg>
/dev/nanda is a normal partition with extFS
<avernos>
uhm, can I edit uboot env directly like that?
<avernos>
oh, but i still need upgrade.. nevermind.. scsi is not on the old uboot i have
<avernos>
not sure what / where to grab and move there. everything within the first partition i guess
<ssvb>
avernos: have you checked the linux-sunxi wiki?
<avernos>
a bit, from there and debian wiki. but im a bit confused.
<ssvb>
avernos: but it refers to already non-existing git branch, and this means that you might need to do a lot of cherry-picking and rebasing yourself, which may or may not be successful
<Supertanker2>
ssvb, no, it seems to be only LWJGL causing the weird display driver bug. I'm poking their devs too to make sure I'm initializing things right.
<Supertanker2>
ssvb, is there any function I need to call like in the raspberry pi (the mysterious bcm_init_host()) or otherwise anything special that needs to be done to get things going?
<ssvb>
Supertanker2: have you checked the high bit of the pointer passed to eglGetDisplay()?
<sunxi_fan>
mripard: in my use case, i'll use both three :-) but i'll force everybody running on a 48KHz rate (using some user level sample rate converter as adaptation layer..)
<Supertanker2>
ssvb, that's on my todo list for today. I was working on this all yesterday and was elated just to have stuff rendering. :)
p1u3sch1 has quit [Ping timeout: 260 seconds]
<Supertanker2>
Ugh, must go wake up. I'll try it later.
<ssvb>
Supertanker2: there is also XInitThreads() issue in the mali blob (you can google for it)
nove has joined #linux-sunxi
<sunxi_fan>
let me clear a bit the patch for:
<sunxi_fan>
1. pcm510x (stub) codec driver (.c and makefile Kconfig)
<sunxi_fan>
2. fix for set parent clock on mod1.c
<sunxi_fan>
3. dts(i) (for enabling the simple card with I2S DAI and codec
<sunxi_fan>
and i'll post for review on sunxi ML; that could be the next iteration.. then we'll see
p1u3sch1 has joined #linux-sunxi
<Supertanker2>
ssvb, interesting, I'll take a look. Thanks!
vagrantc has joined #linux-sunxi
fredy has quit [Ping timeout: 255 seconds]
_massi has quit [Ping timeout: 272 seconds]
engideavr has quit [Quit: Konversation terminated!]
<mripard>
sunxi_fan: that's still very fragile
<mripard>
sunxi_fan: and posting it only to the sunxi ML is a bit pointless
<mripard>
since no-one that will merge the patches reads that ML
fredy has joined #linux-sunxi
<ssvb>
wens, jemk, Tartarus: have you applied for a "free" pine64 development board?
<Tartarus>
ssvb: free? Nope, but I did kickstarter it
<Tartarus>
and I poked the forums about wanting to make sure u-boot works on it
naobsd has quit [Quit: naobsd]
_massi has joined #linux-sunxi
<Supertanker2>
ssvb, yep. It only works when the high bit of display_id is not 1. Looks like it's that bug. :(
<Supertanker2>
ssvb, I'm new to dealing with EGL and X11 explicitly. What is the source of the display ID?
<Supertanker2>
Why does it change with every app invocation?
<ssvb>
Supertanker2: thanks for confirming this!
<ssvb>
it is just an arbitrary pointer, for the simplicity sake, you can consider it a 32-bit random number
<ssvb>
and the mali blob is just playing russian roulette with its high bit
<Supertanker2>
ssvb, hmm. The es2 tri demo works every time. Trying to get it to spit out the display id now so I can see what it's doing.
<ssvb>
anyway, we may apply a binary patch to the mali blobs, but this is a legal gray area
<Supertanker2>
Yeah, I understand.
<ssvb>
on the other hand, the mali developers seem to have confirmed this problem, so this fix is kinda "blessed" by them :-)
<Supertanker2>
ssvb, seems like getting the new blobs is hard, I can't find anything for the Mali400 "utguard" line on their site and allwinner has nothing.
<Supertanker2>
(mali does have a new blob for the 450 it looks like)
<ssvb>
unfortunately new is not always better, I suspect that there are tons of new bugs in the newer mali blobs
khuey|away is now known as khuey
<Supertanker2>
Probably.
<Supertanker2>
ssvb, how interesting. The es2tri demo always has a "non negative" pointer (no high bit set) from XOpenDisplay
<ssvb>
well, it's a smaller program, so its address space allocation may differ
<Supertanker2>
Ah, true that.
<Supertanker2>
LWJGL is, well, Java. So.
<Supertanker2>
Java+JNI stuff for the native EGL.
<Supertanker2>
Still, I have a backtrace, gonna see if there's anything special about how they're getting it.
<ssvb>
the best solution to this whole mess would be of course the open source lima driver, but there are no big corporations to sponsor this work
<ssvb>
we may also try to talk to the allwinner devboard manufacturers and ask them for better mali blobs (they can forward this request up in the support chain)
<Supertanker2>
How far along is the lima driver? I came across it but I'm very bandwidth limited so I didn't try it.
<Supertanker2>
Hmm, so ask cubieboard devs?
<Supertanker2>
Such a cool little device, shame the video driver is bugging like this :(
<ssvb>
yes, in fact the current blobs originate from cubietech
<ssvb>
(as far as I know)
<Supertanker2>
ahh, okay I'll ask them if they have anything newer to try for fun.
<ssvb>
we can also apply the binary patch to the current blobs, and link to the arm support forum in the commit message (just to make it clear that we did not illegally reverse engineer, but only used an "approved" fix)
<Supertanker2>
I'm willing to test it, although I don't have the expertice to come up with the patch myself.
<Supertanker2>
I got two Cubieboard2s for free and I'm not afraid to reinstall.
<Supertanker2>
They were a bit unstable until I tweaked script.bin to up the voltages by 50mv as suggested on their forums.
<Supertanker2>
They were "optimistically low"
clonak has joined #linux-sunxi
<ssvb>
Supertanker2: tweaked voltages compared to what?
<Supertanker2>
ssvb, whatever the Cubian default was.
<Supertanker2>
On a semirelated note, I want to get some heatsinks for the CPUs. I've never liked having bare SoCs, especially on something in a case that actually gets used often (ie compiling stuff)
<ssvb>
did they not release an updated rootfs or something?
<Supertanker2>
ssvb, it wasn't stable enough for me to update from within the device, so I'm not sure. But the release files for cubian have the old script.bin.
_massi has quit [Quit: Leaving]
<Supertanker2>
Other than that, which took a day of debugging, cubian is pretty nice. I'm using it right now. I'm considering going for a month to see if this can replace my desktop for most things (except gaming :P )
<Supertanker2>
I can get down to near where the display is created in LWJGL before magical hand waving JNI starts happening. This is not my field of expertiece lol
<tkaiser>
You'll find them searching for "20x20x5mm Heat Sink For Cubieboard". They're cheap and not that efficient. But at least 10¡C less: http://linux-sunxi.org/User:Tkaiser#Xunlong_settings (50¡C instead of 60¡C with crazy overvolted settings, ambient temperature was around 23¡C)
<Supertanker2>
tkaiser, ahh, I saw a similar one yesterday, I'll give it a try, thanks
<Supertanker2>
-10C is better than no cooling :)
MaDMaLKaway has joined #linux-sunxi
<tkaiser>
If you use a small enclosure I doubt they will be that efficient. I always take care of airflow
<tkaiser>
And A20 on the Cubieboard won't get that hot. Most of the time I've higher AXP209 temperatures.
<Supertanker2>
tkaiser, I leave the little door above the SATA port open, might help a bit, I dunno.
MaDMaLKaV has quit [Ping timeout: 276 seconds]
<tkaiser>
Check before/after: PMU in Cubian should be /sys/devices/platform/sunxi-i2c.0/i2c-0/0-0034/temp1_input
<Supertanker2>
tkaiser, hmm I won't worry about it too much then. Back to tracing byzantine calls through JNI to figure out where this X11PeerInfo struct is being created.
<Supertanker2>
tkaiser, wow that is a beautiful tool (rpimonitor)
afaerber has quit [Quit: Ex-Chat]
<tkaiser>
Used with the old 3.4 kernel it can also give some good insights into consumption since AXP209 exposes nearly everything through sysfs
<Supertanker2>
I am on 3.4.79
avph has joined #linux-sunxi
<Supertanker2>
ssvb, I'll try writing to cubietech and seeing if they have a newer driver. Thanks for the help today :)
<Supertanker2>
(Somebody else's code, not mine, but cool to see still)
<Supertanker2>
ssvb, mind if I write up a blog post about it? Might help some people.
<Supertanker2>
Cubieboards are awesome :o
<ssvb>
libv: would we apply a binary patch to the mali blob in the sunxi-mali-proprietary repository to fix this crap?
<Supertanker2>
I love how it's like 1 byte.
<Supertanker2>
Well, 4 I guess.
<Supertanker2>
What about writing a tool to patch it?
<Supertanker2>
Because that's not disproportional effort.
<Supertanker2>
(Sarcasm)
<Supertanker2>
I wonder what the proper behavior is for the pointer.
<Supertanker2>
Is zero an error, and everything else not an error?
reinforce has quit [Quit: Leaving.]
<ssvb>
Supertanker2: this is some kind of "sanity" check in the mali blob, which rejected NULL pointer and also negative pointers
<ssvb>
Supertanker2: I have changed it to fail only for NULL pointer (return 0) and succeed otherwise (return 1)
<Supertanker2>
ssvb, the negative is hilarious. That's the kind of mistake we talked about endlessly in my computer science courses (I'm a newly minted teaching associate for grad school)
<Supertanker2>
Ooo I can use this for an example in my intro to CS class.
lobinho has quit [Ping timeout: 256 seconds]
Netlynx has quit [Quit: Leaving]
<lamer14525387307>
BTW: Any Pine64 arrived in the meantime?
<Tartarus>
This way I feel less bad when I don't have the time to poke stuff that I wanted
<ssvb>
ok, are you receiving yours from kickstarter on February or at a later date?
<ssvb>
jemk, wens: ^
tkaiser has joined #linux-sunxi
<Supertanker2>
ssvb, what are pine64 boards? New ARM devices?
<WarheadsSE>
pine64.org
<Supertanker2>
Oh wow, that looks really nice.
<WarheadsSE>
meh
mosterta has joined #linux-sunxi
<Supertanker2>
The $15 I mean.
<jelle>
Supertanker2: but it isn't $15 :p
<Supertanker2>
I'm a poor but not quite starving college student so low prices draw me in.
<Supertanker2>
Oh :(
* Supertanker2
sobs
<jelle>
there is additional shipping I think
<Supertanker2>
From where?
Worf has joined #linux-sunxi
<jelle>
seems to be $12 everywhere
<Supertanker2>
My connection is very slow right now, I stopped loading the site. :(
<Supertanker2>
Eh, $30 still isn't bad for a quad core ARM.
<Supertanker2>
Although that's about rPi2, isn't it?
<Supertanker2>
Ah well, choices are good.
<WarheadsSE>
+ shipping, then other needs (sd card, power supply, etc) and the complete lack of mainline as of yet
<WarheadsSE>
You know "ridiculously great deal" .. in a year
<Supertanker2>
lol
<WarheadsSE>
Had they given us SATA, pfft
<Supertanker2>
I found out that, in my hoard of parts, I happened to have two 5V 2.1A power bricks for the two Cubbies I got.
<WarheadsSE>
I have one from them sitting at my desk at home. Not even going to touch it with the lichee pile.
<Supertanker2>
So I don't have to use my phone charger anymore.
<jemk>
ssvb: i have applied, but only recently, somehow missed the mail. so i haven't hear from them yet
<tkaiser>
Supertanker2: Xunlong currently ships their $15 H3 based Orange Pi Pc for $3.5 worldwide (and fast also). And I doubt A64 will be that faster than H3
<Supertanker2>
tkaiser, a friend has one, he says it's pretty cool now that they have proper hw graphics acceleration (?).
<Supertanker2>
Haven't looked into it much, but I should get one.
<Supertanker2>
I have a few Intel Galileo boards too but don't know what to do with a 400MHz CPU.
<jelle>
Supertanker2: create a cluster obviously :)
<Supertanker2>
jelle, actually, at the last intel dev forum I went to, a guy had 40 Galileos stacked to make a cluster
<Supertanker2>
I was envious. Galileos are kinda expensive for what they do though so it was probably $1-2k USD worth of hardware at least
<Supertanker2>
The heat rising from the cluster was duly impressive.
<WarheadsSE>
lol jelle
<Supertanker2>
I forgot what he was clustering for.
<Supertanker2>
I think he was testing a distributed hash table algorithm.
<tkaiser>
WarheadsSE: IIRC you received your Pine64 dev sample already?
<Supertanker2>
I should actually buy a rPi, all my friends have one, clearly I should too.
<Supertanker2>
11!!11one
domidumont has quit [Ping timeout: 246 seconds]
<WarheadsSE>
tkaiser: that's what I just said, essentially
<WarheadsSE>
Got in before the weekend
<tkaiser>
Ok, so you're the man able to do the FEL mode test ;)
<WarheadsSE>
not at the moment, no
<WarheadsSE>
but in theory, yes
<tkaiser>
I asked TL Lim 4 weeks ago and ssvb yesterday. But it seems the Pine people do exactly nothing to help here.
<WarheadsSE>
hah
<WarheadsSE>
I can attempt at some point, I will be from home tomorrow, which gives me a small amount
<tkaiser>
I would believe ssvb is interested in results. I'm already too scared by their communications
<Supertanker2>
lol
<tkaiser>
And I don't get it why they do not manage to ship developer samples fast
<tkaiser>
I got two boards sent from Shenzhen (LeMaker and SinoVoip) and it took just 48 hours for the stuff to be sent to Germany
mosterta has quit [Ping timeout: 255 seconds]
soderstrom has quit [Ping timeout: 272 seconds]
<Supertanker2>
Hmm is there a way to give more memory to the video chip on the cubieboard? I'm not sure what settings to tweak.
<Supertanker2>
Instinct says it's the fb memory
<yann|work>
trying to build linux-sunxi 4.4 (for OrangePi-Plus), I'm hit by the "multiple (or no) load addresses" issue - are we supposed to specify a load address, or to switch to zImage ?
<ssvb>
Supertanker2: are you running out of memory?
<Supertanker2>
ssvb, I don't know. I'm trying to tweak a bit more performance.
<Supertanker2>
ssvb, do you know if the r3p0 blob enables both cores in the Mali chip?
<Supertanker2>
I read somewhere it has 1 vertex and 2 fragment processing units.
<ssvb>
Supertanker2: yes, the r3p0 blob only uses 1 fragment processing unit, the r3p2 can use both
<Supertanker2>
ssvb, how can I switch to r3p2?
<Supertanker2>
That way I can test the other patched blob too :)
<Supertanker2>
Do I need to rebuild libump, the kernel driver from sunxi-mali, etc ?
<ssvb>
Supertanker2: you need a kernel with r3p2 mali driver version
<Supertanker2>
Ahh.
<Supertanker2>
That might be a bit much for me to tackle today with a low bandwidth connection.
<Supertanker2>
I'll jot it down as future work. Thanks!
<Worf>
sigh ... what have i gotten myself into? Just wanted to peek what is out there (beyond the RPi) and now i'm knee deep in reading about issues i'm not sure i wanted to know about. And it's to late to run away.
<yann|work>
hm, UBOOT_LOADADDRESS in conf/machine/include/sunxi.inc looks like a real typo... is there even a need to set it, given the default in kernel.bbclass ?
<yann|work>
Worf: what are you playing with ?
<Worf>
yann|work: waiting for a RPi and a Pine64+ to arrive ... the 2nd obviousely made me check out this place
<ssvb>
Worf: what are you expecting from these boards?
<ssvb>
with Pine64+ you will likely first get some sort of Ubuntu (or Android) running on a 3.10 kernel
<KotCzarny>
or just promises
Gerwin_J has quit [Read error: Connection reset by peer]
<Worf>
for now: experience. probably i'll get some more alternatives. especially if i can require some funding from my workplace. we will see.
<ssvb>
KotCzarny: getting Ubuntu running on a vendor provided kernel is not a rocket science
Gerwin_J has joined #linux-sunxi
<vagrantc>
depends on the vendor...
<KotCzarny>
ssvb, we will see
ricardocrudo has quit [Ping timeout: 240 seconds]
<ssvb>
vagrantc: did you have a really bad experience with some vendor kernel?
<tkaiser>
KotCzarny: If the kernel boots then we'll see a flood of crappy Linux OS images coming over Pine64
<tkaiser>
I believe this will be the main problem for the Pine guys or their backers
<vagrantc>
ssvb: i've seen some that kernel panic if you try to do anything more than idling.
Supertanker2 has quit [Read error: Connection reset by peer]
<vagrantc>
not exactly what i'd call a good experience... :)
<ssvb>
tkaiser: my understanding is that Pine64 people are currently negotiating with the Remix OS guys
<tkaiser>
ssvb: Sure
<tkaiser>
But that would only improve the Android experience?
<tkaiser>
And I don't get how the Remix folk might profit from this deal. Their business modell is to sell Remix OS to hardware vendors
bonbons has joined #linux-sunxi
<ssvb>
probably just in the same way as Microsoft profits from the Raspberry Pi :-)
<ssvb>
that's a way to show to the whole world that you are not dead yet
<KotCzarny>
and install own spyware?
<tkaiser>
Agreed. But again: Linux
<ssvb>
what kind of software support did they promise on the kickstarter page?
<tkaiser>
If I understand correctly then their BSP with kernel 3.10.65 means booting into an initrd and you would need to have a virtual machine with their BSP to change even most basic settings
<tkaiser>
ssvb: You're right: Nothing except of Android and also a rather weird Linux kernel release strategy (3.18 in 3 months and 4.2 in 6 months)
<tkaiser>
That scares me a bit
<ssvb>
tkaiser: how is this initrd approach different from what they offer in the H3 SDK?
<tkaiser>
There's no difference
Supertanker2 has joined #linux-sunxi
<ssvb>
one can just disable this initrd stuff and boot the kernel normally (at least on H3)
<tkaiser>
But users now expect something like 'script.bin' to adjust settings
<tkaiser>
True. But look (better not) at the situation with Banane Pi M2/M3.
<tkaiser>
These guys still don't get it to escape from initrd. And the SinoVoip people have at least some basic knowledge about this. The pine guys seem not be interested to get to this point
<KotCzarny>
so, basically, just another cheap tablet
<tkaiser>
Therefore I fear that they will be flooded with countless combinations of u-boot+kernel+initrd + some rootfs all being more or less crappy
<KotCzarny>
with android
<tkaiser>
Leading to a horrible 'Linux experience'
<ssvb>
they just need some basic QA and an 'offical' Ubuntu image
<tkaiser>
You're totally right. But at the moment I fear they realise this
<tkaiser>
Based on the offerings to every weird people appearing in their forums and claiming 'I've already combined A with B. I will do the job to provide xxx support!'
premoboss has joined #linux-sunxi
mawe242 has quit [Quit: Leaving]
* yann|work
tries to understand how the yocto linux-sunxi tree manages to build a uImage for jethro
iamfrankenstein1 has joined #linux-sunxi
<yann|work>
looks like the UBOOT_LOADADDRESS typo got unnoticed only because it's not used...
<Worf>
well, i already figured that there seems to be no device (at least not in the RPi price area) that you can use without proper medication. but luckily companies like nvidia did prepare me well...
<yann|work>
Worf: that's not completely untrue, I must say :)
<yann|work>
even the RPi under Raspbian needs some work (eg. not shipping a rpi-specific libsdl2)
<yann|work>
I have a couple of yocto patches for BPi-M1, but that one is so a huge a mess that i'm not sure I'll want to finish it - especially as I have more powerful cards at hand now
afaerber has joined #linux-sunxi
<ssvb>
Worf: it depends on what you need, older Allwinner devices work fine
<ssvb>
newer devices with new SoC variants, such as H3 and A64 need some work to get fully supported in the mainline kernel
<ssvb>
if you are interested in Android, then Allwinner tablets work mostly fine out of the box with all the games, multimedia, netflix, skype, etc.
<ssvb>
of course, Pine64+ is very new, so people are worried that the initial software support may or may not be somewhat limited for GNU/Linux
<Supertanker2>
Those HDMI connectors make me nervous on the cubbieboards
<Supertanker2>
They're very tight and they look fragile
<yann|work>
ssvb: btw, H3 is supported by linux-sunxi, right ? Maybe just on sunxi-next ?
<Worf>
well, my interests are covering a broad spectrum. and i think i understand that there is not one single device that can handle all that. and i'm prepared to not get very far with linux on the Pine64+ for a while.
Supertanker2 has quit [Read error: Connection reset by peer]
<yann|work>
but even if the older SoCs are supported by kernel and u-boot, no distro today provide an out-of-the-box experience to take advantage of the hardware, there's a lot of DIY left to the user, finding the correct software, compiling, etc
<yann|work>
ssvb: ok thx
tkaiser has quit [Ping timeout: 245 seconds]
<Worf>
yann|work: any device reasonable supported and ... "capable" worth recommending? (i think i did read a lot "orange" by now)
<ssvb>
yann|work: if you don't mind running on the Allwinner's 3.4 kernel, then you can have everything including graphics acceleration and hardware video decoding on A10/A20
Supertanker2 has joined #linux-sunxi
<ssvb>
yann|work: if you are interested in a server system, then you can have the mainline kernel running on these old SoC's too (with SATA, USB, Ethernet and everything else working)
<yann|work>
ssvb: as I read the page, it does not say anything about 4.4, and for 3.4 just about using a lichee tree
<yann|work>
ssvb: I'm more interested in the video stuff
<yann|work>
Worf: hard to say for now, and depends on what you want to do with the board - my experience is centered around video (h.264 decoding and rendering)
<ssvb>
yann|work: there is no support for H3 in the 4.4 kernel, some H3 support is expected to be added in 4.5
<ssvb>
yann|work: is the "Mainline kernel" a confusing term for you?
<Worf>
yann|work: for non-video centered i think i know what to do. recommendations where you have experience is welcome
<yann|work>
ssvb: np I'm familiar with "Mainline kernel" :)
<yann|work>
I'm talking about the 4.4-based sunxi-next branch, which I expected to maybe contain H3 code (especially as it contains defconfigs for many H3 boards)
IgorPec has quit [Ping timeout: 246 seconds]
iamfrankenstein1 has quit [Ping timeout: 246 seconds]
<ssvb>
yann|work: yes, you can use it at your own risk
nemunaire_ has joined #linux-sunxi
<yann|work>
Worf: the RPi is quite easy to work with (only had to rebuild sdl2), but is not a very powerful board - you can easily get things to work using OpenMAX (or rather the non-standard bcm-specific ilclient waper) ; but then the AllWinner boards have poor OpenMAX support (just decoded, no renderer), so even porting ilclient to them is likely useless, vdpau is the way to go there, it seems
<ssvb>
yann|work: you can also say that RPi has a poor VDPAU support, it depends on the perspective
<yann|work>
but then, the official Raspbian for BPi-M1 is a joke (kernel USB stack exploding on device-plug), so I've turned to yocto, but there very view graphics software has been integrated properly for now - I have some patches but nothing complete yet
<yann|work>
ssvb: I only suspected that, did not get any first-hand experience with it
SadSmile has joined #linux-sunxi
tkaiser has joined #linux-sunxi
<zerotri>
for those with experience with the different mali blobs: are these generally board agnostic or do you need to make sure they are built to support multiple boards?
SadSmile has quit [Quit: Leaving]
nemunaire_ is now known as nemunaire
viccuad has quit [Ping timeout: 260 seconds]
apritzel has joined #linux-sunxi
khuey is now known as khuey|away
<yann|work>
ssvb: at least the sdcard I generated does not appear to boot much of the board - if only I had not left my debug cable at the office... :) have to go to sleep for lack of tools, isn't that a shame...
<yann|work>
btw, the reason why I've started to play with building the kernel, is the lack of mali driver in the kernel H3 kernel built by loboris
iamfrankenstein1 has joined #linux-sunxi
<yann|work>
Worf: btw, I forgot the "official raspbian" for OrangePi-Plus is to be avoided as well (X11 display not even working properly), have to find other images, with the ones form loboris being quite decent, just without any hw-specific stuff for graphics
<yann|work>
in fact I'm not sure I'll find the mali drivers in sunxi-next either anyway...
<yann|work>
ssvb: I admit not looking that the source yet :)
<vagrantc>
any folks around worked with cubieboard4? i need a recent version of u-boot, mainline u-boot looks to have A80 support, but not cubieboard4 ...
<vagrantc>
or even experience using a vendor u-boot with a mainline kernel...
ricardocrudo has joined #linux-sunxi
<yann|work>
ow, a 2011.09 u-boot :)
* vagrantc
nods
* vagrantc
is often surprised to see really new boards using exceptionally ancient versions of u-boot