<hno>
Heh.. been pulling my hair for a while trying to figure out why JTAG crashes after a while in the merged sunxi tree, only to realize that it's the watchdog I added myself that kicks in.
<hno>
u-boot that is.
<Triffid_Hunter>
heh
<mnemoc>
hno: :D
<Triffid_Hunter>
whenever I use watchdog with stuff, I make it print out during reboot that the watchdog triggered last time.. not sure how easy it is to do that with these arm chips though
<hno>
RaYmAn or mnemoc, do you have time to test u-boot from the sunxi branch on both A10 and A13?
<mnemoc>
hno: I'll prepare my desk for that now
<hno>
Triffid_Hunter, yes that would be nice, but I have no idea to tell why these Allwinner chips reset.
<hno>
and to complicate further I had no console cable connected, only JTAG.
<cheng>
hno: i have console cable connect, would be good if i can helps out
<hno>
cheng, sure, just grab the sunxi u-boot branch, build for sun4i or sun5i depending on what you have and prepare an SD card with the result and try booting.
<hno>
If all is well you should see SPL banner, and then u-boot starting.
<RaYmAn>
I'm not at home (and devices are at home), so can't help unfortunately
<hno>
I have only been able to test A13 SPL via JTAG so far. And now at work and can't do any testing until late tonight.
<cheng>
i'm still in learning stage. previously i have clone uboot-allwinner, so now i just issue 'git pull', what else i need to do before building?
<hno>
you need to switch to sunxi branch. Default branch is still sun4i.
<hno>
git checkout -b sunxi origin/sunxi
hipboi has joined #arm-netbook
<cheng>
i believe compile command still this 'make ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabi- sun4i', right?
<lundman>
i just say make, but then I compile on arm
<hno>
need to know if that's the spl build or u-boot build.
popolon has joined #arm-netbook
ibrah2 has quit [Ping timeout: 272 seconds]
polto2010 has quit [Ping timeout: 272 seconds]
tzafrir has quit [Ping timeout: 240 seconds]
ibrah2 has joined #arm-netbook
sspiff has quit [Remote host closed the connection]
sspiff has joined #arm-netbook
mysteryname has joined #arm-netbook
von_fritz has joined #arm-netbook
<oliv3r>
lkcl: but there are hopes then? E.g. there might be reason to invest into a a10 based device? It really sucks that AMLogic didn't want to play ball, as there atleast is some software available for their vpu :(
<von_fritz>
hno Loading U-Boot... ### ERROR ### Please RESET the board ### nohting more on mele
hipboi_ has joined #arm-netbook
<oliv3r>
I wonder how well an A10 based native (or i guess even with android) XBMC box would sell. If enough momentum could be gained so a big enough order could be generated, that would be a nice incentive. But with all the android xbmc boxes out now, it's highly unlikly people would be interested :(
ibrah2 has quit [Ping timeout: 246 seconds]
hipboi has quit [Ping timeout: 240 seconds]
<hno>
von_fritz, nothing else? Ok. will look into it tonight. Thanks for testing.
ibrah2 has joined #arm-netbook
<rz2k>
hno: ### ERROR ### Please RESET the board ### is the error I've got on my ly-f1 when it was working, fixed with board soft reset by onboard button wired to A10 reset. never did boot up from cold start.
<rz2k>
s/got/had/
<ibot>
rz2k meant: hno: ### ERROR ### Please RESET the board ### is the error I've had on my ly-f1 when it was working, fixed with board soft reset by onboard button wired to A10 reset. never did boot up from cold start.
ibrah has joined #arm-netbook
<hno>
rz2k, it's a fairly generic error. Error reporting in SPL isn't the best.
<mnemoc>
oliv3r: you can configure xmbc to use an external player, like mx, which then uses the native cedarx support in android 4.0
<mnemoc>
oliv3r: not nice, but at least better than what we have on linux
<mnemoc>
hno: building sun5i, make[2]: *** No rule to make target `/home/amery/projects/allwinner/hno-uboot-sun4i/include/linux/types.h', needed by `cache_v7.o'. Stop. ... should I copy it or symlink something?
<oliv3r>
Well, as a 'user'; I much support what rhombus tech aims and wants to do, but atm I think, the most 'open source' we have right now, is the pivos (amlogic) I haven't played much with the source yet though and the mediaplayer sure does look very messy
<hno>
mnemoc, did you remember CROSS_COMPILE?
<mnemoc>
hno: yup
<hno>
mnemoc, try "distclean"
<oliv3r>
anybody know btw, what OpenMAX goal is? Is it to 'replace' vpdau/XvBA/VA-API/xvmc etc? be 'the one to rule them all' kinda?
<mnemoc>
hno: meh. distclean did the trick.... a mere `clean` wasn't enough
<mnemoc>
odd
ibrah2 has quit [Ping timeout: 246 seconds]
<mnemoc>
hno: does uboot support an O= like argument to be able to build both over the same working tree?
pawel58701 has joined #arm-netbook
pawel5870 has quit [Quit: Leaving.]
<mnemoc>
it does :)
<hno>
mnemoc, no idea. I hope so.
<mnemoc>
$ du -sh sun*
<mnemoc>
6,8Msun4i_build
<mnemoc>
6,8Msun5i_build
<mnemoc>
:)
<mnemoc>
now to test
ibrah2 has joined #arm-netbook
<hno>
mnemoc, build worked now? What was the cause to types.h error?
hipboi_ has quit [Read error: Connection reset by peer]
mysteryname has quit [Quit: mysteryname]
<mnemoc>
hno: I had renamed the working dir from hno-uboot-sun4i to hno-uboot-sunxi but that detail was important to make, so distclean solved it
<mnemoc>
absolute references are anoying :<
<mnemoc>
but more annoying are my ttl/usb adapters... they like to go on strike :<
ibrah2 has quit [Ping timeout: 252 seconds]
<mnemoc>
ordered a bunch more so I don't have to move them from one device to the next
tzafrir has joined #arm-netbook
<cat_x301>
mnemoc: what is the recommended place to download rtl8192cufw.bin?
<mnemoc>
ow
<mnemoc>
/lib/firmware/ is the standard place
<mnemoc>
I had forgotten the mainlined wifi driver needs that
<cat_x301>
mnemoc: to download from actually :)
<mnemoc>
your desktop machine? :p
<mnemoc>
want my /lib/firmware/rtlwifi/rtl8192cufw.bin ?
<cat_x301>
hmm, there is something in gentoo too, not sure hoever, because last two letters are different: net-wireless/rtl8192su-firmware
<mnemoc>
in rtlwifi I have c, cu, de, se and u. but not su
<cat_x301>
:)
<cat_x301>
anyway, got one from your link, thanks!
<mnemoc>
I suppose it might be a good idea to include this .bin in our tree if we are going to use that driver
<mnemoc>
yw
mysteryname has joined #arm-netbook
ibrah has quit [Ping timeout: 240 seconds]
<cat_x301>
mnemoc: affirmative
<specing>
cat_x301: Another Gentoo user? LOVELY
<specing>
cat_x301: join #gentoo-embedded
<specing>
cat_x301: Everything should be in linux-firmware
<cat_x301>
specing: actually i am not that advanced with it. Rarely do any update, hardly understand packaging system -- in short use it like a monkey :) but like it anyway!
<mnemoc>
not sure how the GPL license of that code affects the blob
<cat_x301>
mnemoc: ah, you're right, i was talking about upstream driver. orly_owl: sorry
pawel5870 has joined #arm-netbook
<mnemoc>
eventually we all should use the upstream driver
* cat_x301
: mer does not have rfkill tool :(
<orly_owl>
mnemoc: so you can view the firmware source code or it's in binary form?
<mnemoc>
disable rfkill in your kernel build
<cat_x301>
mnemoc: brilliant idea!
<mnemoc>
orly_owl: binary, as a char[] :p
<orly_owl>
darn
<orly_owl>
ok
<cat_x301>
i guess one can easily make firmware blob out of it.
<hno>
mnemoc, any success/failure in u-boot tests?
<cheng>
hi hno. i just tried with the u-boot, boot ok.
<cheng>
but once i change to to sun4i, it no longer work.
<hno>
cheng, which one works, and which fails?
<cheng>
sunxi working..
<cheng>
sun4i not working
<hno>
what board do you have?
<lundman>
finally I can port mirror and do some packet sniffing
<cheng>
mele a1000
<hno>
sun4i branch should work fine on a1000.
<mysteryname>
cheng: does yours hang? IIRC I tried sun4i on the A2000 and it hung, so I have reverted it till I have more time to play around.
<orly_owl>
make firmware blob out of it? meaning?
<cat_x301>
orly_owl: no sense as long as there is working upstream driver :)
jlj has joined #arm-netbook
Almamuetya has joined #arm-netbook
<orly_owl>
oh
<orly_owl>
upstream driver doesnt load firmware onto the wifi chipset?
<hno>
orly_owl, it does in most cases.
<hno>
from a file.
pawel5870 has quit [Remote host closed the connection]
<orly_owl>
ok
<orly_owl>
is there a list of wifi chipsets a10 tablets use?
<hno>
almost no upstream drivers have firmware embedded in the driver.
<hno>
orly_owl, most are using various realtek modules.
<orly_owl>
ok
<hno>
Allwinner have a list they support.
<hno>
why?
<cheng>
for sun4i uboot, kernel stop at cannot open root device "mmcblk0p2"
<orly_owl>
all of which either load firmware using the driver or have firmware embedded in the driver
<orly_owl>
id like a tablet that doesnt require loading firmware onto the wifi chipset
pawel5870 has joined #arm-netbook
<hno>
why?
<hno>
cheng, make sure there is rootwait on the kernel command line.
<orly_owl>
for the freedom aspect
<orly_owl>
free software and so on
<hno>
cheng, so both sunxi and sun4i branches suceed in loading the kernel for you, but with sun4i the kernel fails to mount the root?
<cheng>
yes, both sun4i and sunxi can load the kernel.
<hno>
orly_owl, so you are of the opinion that a firmware rom on the wifi module is more free than a binary file with the same contents loaded runtime?
<hno>
well, both are loaded runtime.
<hno>
only differing in by whom and from where.
<mnemoc>
hno: haven't been able to test yet :<
<orly_owl>
hno: yes, im funny like that
<merbanan>
oliv3r: not completely but it is supposed to be a generic interface against codecs, but it isn't
<cheng>
yup, 'rootwait' solve it..now ok
<cheng>
but in all cases, i do not use the spl
<orly_owl>
i consider firmware burnt onto a rom part of the hardware
<RaYmAn>
orly_owl: almost all, if not all, wifi chips used in embedded devices have binary firmware that is uploaded runtime. (BCM4329/4330 is one, very popular, example)
<cat_x301>
orly_owl: in that case you should not worry about how it ends up there in hw
<orly_owl>
RaYmAn: mhm, thats why i prefer atheros wifi
<orly_owl>
for laptops
<orly_owl>
obviously not really an option for a10 devices
<RaYmAn>
iirc atheros embedded chipsets also loads firmware runtime.
<mnemoc>
orly_owl: eoma68-a10 + good-laptop with atheros wifi ;-)
<orly_owl>
RaYmAn: nah, it stays on the wifi card
<orly_owl>
mnemoc: ok, let me know when eoma68-a10 is for sale :3
<mnemoc>
orly_owl: i believe before xmas... but we still need someone to make a good eoma68 laptop :<
<orly_owl>
there is an ath6k?
<orly_owl>
mnemoc: until then, time to gut existing laptops and stick an eoma68-a10 inside
<RaYmAn>
well, the driver is called that. I guess it's more AR6xxx devices.
<oliv3r>
merbanan: so it's doomed to fail? :p
<mnemoc>
orly_owl: :)
<orly_owl>
or coreboot my thinkpad
<orly_owl>
Either way
<merbanan>
oliv3r: no, android uses omx, and the spec has been revised
<merbanan>
but it is a mess
<merbanan>
you can't implement omx support for one driver and have it work for another
<merbanan>
there are quirks in all different implementations
<RaYmAn>
It's still a hell of a lot better than people using their own proprietary stuff (but definitely; it should be much more standardized)
<oliv3r>
ah, I thought OMX was OpenMaX :p
<oliv3r>
merbanan: so what would you think the best long term solution for hw assisted decoding is? something like vpdau/xvba/cedarx? something more generic suitable for all? OpenCL/Max/OMX?
IEF has quit [Quit: brb]
<merbanan>
omx is openmax
<merbanan>
I guess a better specified omx is the way to go
<merbanan>
but right now it is a mess
<merbanan>
just check all the quirk flags the stagefright omx layer has
IEF has joined #arm-netbook
<oliv3r>
and does the khronosgroup have any plans for that?
<oliv3r>
personally, I guess opencl would be a better solution, and simply have proper backends for neon and maybe even dedicated 'video' units
<mnemoc>
libav ftw!
<mnemoc>
err... libva
cheng has left #arm-netbook ["Leaving"]
popolon has quit [Read error: Operation timed out]
* cat_x301
still has no luck with wifi..
<cat_x301>
kernel bug:
<cat_x301>
[root@localhost ~]# ifconfig wlan0 up
<cat_x301>
[ 81.970000] rtl8192cu: MAC auto ON okay!
sspiff has quit [Remote host closed the connection]
<hno>
cat_x301, high quality driver code. There should be more info in the oops pinpointing more exactly where the error is.
<cat_x301>
hno: yeah, i know, but it is a bit inconvenient to post duplicated messages here..
<RaYmAn>
cat_x301: pastebin ;)
<cat_x301>
RaYmAn: no only that, i have to remove every second line by hand to provide some sane reading :)
<cat_x301>
s/no/not
<RaYmAn>
oh yeah - Disable low-level debug output. If you boot to regular console, there's not much point in that
<cat_x301>
i asked already about getting rid of messaging like this:
<cat_x301>
<6>rtl8192cu: MAC auto ON okay!
<cat_x301>
[ 81.970000] rtl8192cu: MAC auto ON okay!
<cat_x301>
it does not seem to be that low level.
<cat_x301>
priority 6..
<RaYmAn>
the <6> messages are the low-level debug thingy
<RaYmAn>
it's an allwinner thing.
<cat_x301>
RaYmAn: how come 3.6 kernel does not do anything like this, with the same codebase+defconfig?
<cat_x301>
well, allwinner codebase :)
<RaYmAn>
you probably didn't add in the debug code :)
<cat_x301>
RaYmAn: :)
<cat_x301>
but hey, i am using upstream version of rtlwifi
<cat_x301>
no allwinner cr#p
<RaYmAn>
fine, let me expand on my statement. Allwinner registers a serial debug console very early in their code. (IIrc, hardcoded somewhere outside of anything allwinner should ever touch)
<oliv3r>
does the a10 allready run a 'mainline' kernel?
<cat_x301>
oliv3r: at least i get up to console.
<RaYmAn>
cat_x301: hence, all messages written to console are put out on that debug console. Then when the 'regular' console registers, double messages appears.
<cat_x301>
RaYmAn: to me it looks like stderr and stdout are simply printed on the same console.
<RaYmAn>
cat_x301: I give up.
<cat_x301>
RaYmAn: please don't! :)
<grevaillot>
hi all
<grevaillot>
dupplicated log is not an allwinner thing, it's normal when using LL_debug with a regular console on the same uart
<oliv3r>
or rather, a 'vanilla' kernel?
<grevaillot>
just config issue :)
<hno>
RaYmAn, maybe they did not know about earlyprintk?
<cat_x301>
grevaillot: what is config option to get rid of this? early printk does not affect
<RaYmAn>
hno: I guess so :P
<hno>
cat_x301, getting your hands dirty in the kernel code and clean up the mes.
<hno>
s
<cat_x301>
RaYmAn: i think you are right, look:
<cat_x301>
[ 0.430000] sunxi-uart.0: ttyS0 at MMIO 0x1c28000 (irq = 1) is a U6_16550A
<cat_x301>
[ 1.110000] console [ttyS0] enabled
<cat_x301>
it starts duplication exactly after this line
<cat_x301>
<6>console [ttyS0] enabled
<grevaillot>
hmm, they might have their early_printk stuff not plugged n ;)
* grevaillot
total allwinner noob...
<grevaillot>
alinuxino a13 in the mailbox yesterday
<RaYmAn>
cat_x301: gee, that almost sounds like it's saying I'm right ;)
* hno
still remembers running 3.4 with only earlyprintk console, no ttyS0. Was a bit limiting.
<RaYmAn>
kind of one-way, isn't it? :P
<hno>
grevaillot, good timing just as we got u-boot working decent.
hipboi has joined #arm-netbook
<hno>
RaYmAn, it is.
<RaYmAn>
I wish I could remember where allwinners earlyprinkt-type stuff is hidden
jinzo_ has joined #arm-netbook
popolon has joined #arm-netbook
<grevaillot>
hno: i was looking at your uboot tree, not sure of what it implies - discovering the boot chain process on sunxi platform
<hno>
I am a little confused that it has been allowed to surive.
<hno>
grevaillot, what?
<grevaillot>
hno: are we not dependent anymore of allwinder boot0 and boot1 blobs ?
<hno>
grevaillot, we have not depended on them in quite a while, but only on A10. Now on A13 as well.
<grevaillot>
ah, nice
<hno>
but only for SD booting so far. Sill have to crack some details about the nand contorller/driver.,
<jinzo_>
Hello, I'm looking around and see a boatload of A10 + Mali400 boards/devices, but does anyone know if there're any without Mali400 (and with SATA)? So something aimed at small servers/alike?
<RaYmAn>
mali is inside the a10 SoC
<hno>
jinzo_, MALI is always there. but you don't need to use it if you don't want it.
<hipboi>
jinzo_: checkout cubieboard.org
<hno>
oh, hi hipboi
<jinzo_>
that's what I feared. hipboi I'm having my eye on it
<hno>
jinzo_, why fear?
<hipboi>
hno: hi
<jinzo_>
hno, because I hoped there will be some even cheaper devices without a GPU available sometime
<hno>
jinzo_, the A10 isn't exactly expensive.
<jinzo_>
indeed - but why would I pay for something I won't need?
<Marex>
hno: hey
<Marex>
hno: how're things going ?
<mysteryname>
jinzo_, it's like asking for a intel with out the HD4000 in it, I don't think that can be done.
<hipboi>
hno: have you got a uSD breakout board
<jinzo_>
but, looks like I'll have to snag cubieboard and play with that :)
<hno>
Marex, quite well. Merging sun4i & sun5i trees, and then cleaning the tree up for submission.
<hno>
hipboi, no, was about to ask you what happened with that
<oliv3r>
Most servers have integrated graphics, but not loading the driver is usually enough to not waste any resources or increase temperature :)
<Marex>
hno: perfect
<Marex>
hno: thanks!
<hipboi>
hno: because i said i will send you one
<hipboi>
hno: with the cubieboard
<Marex>
hno: I'm now doing a gigantic stdio/serial rework
<jinzo_>
oliv3r, indeed - and probably easier to upgrade/hack if you don't relay on proprietary GPU drivers :)
<hno>
hipboi, yes...
* hno
don't have a cubieboard either.
<hipboi>
hno: will start to ship this weekend, hopefully
<oliv3r>
and that :) as for price, I guess you could find a SoC without a GPU in it but doubt it'll be any cheaper.
<jinzo_>
hipboi, regarding your shop - ever considered selling RF (presuming the Mele F10 sensor remote is RF) usb dongles?
<jinzo_>
because I found out that it's hard to get a good "generic" remote.
<hno>
The Mele remotes include an RF dongle from what I can remember.
<rz2k>
hipboi: any news about mali r3p1?
<jinzo_>
hno, oh? Then I missed it. That would be great.
<hipboi>
cubieboard is too small to put another usb hub chip and a rf module
<hipboi>
there is ir on the board
<hno>
hipboi, you have USB ports. That's sufficient.
<jinzo_>
hipboi, I'm not talking about cubie itself - but more as a generic solution I could plug in
<oliv3r>
hipboi, about your shop, why is shipping to europe so horrendosly expensive? I was looking at a mele a1000 but 75USD for the device, 42USD for shipping is ... harsh :)
<jinzo_>
to any of the myriad of A1X devices.
<jinzo_>
oliv3r, that's quite standard for aliexpress and DHL/others.
<jinzo_>
the cubieboard shipping is 29USD tho :D
<oliv3r>
yeah, but when I order via dx; i get the same device, with "free" shipping
<mnemoc>
jinzo_: pick china post, not fancy carriers
<oliv3r>
I don't mind spending a little extra as sponsoring
<jinzo_>
yeah, with China Post (so slower)
<mnemoc>
2w max
<jinzo_>
mnemoc, yeah I know - but I think I didn't notice china post on much of aliexpress listings
<oliv3r>
It didn't seem like it was changable
<mnemoc>
jinzo_: what country?
<Triffid_Hunter>
oliv3r: I ordered one from DX.. 6 weeks later I cancelled the order, they never shipped and never gave any reason why not
<jinzo_>
mnemoc, Slovenia
<jinzo_>
Triffid_Hunter, probabbly means only shipping
<oliv3r>
I've orderd many many things via dx
<jinzo_>
so from the time they ship to the time you recieve it, and it's somewhere there - from 2 to 3w to europe
<mnemoc>
jinzo_: send hipboi a message asking what country to pick instead
<jinzo_>
mnemoc, indeed - Austira (close to us) offers china post
<mnemoc>
jinzo_: aliexpress doesn't let sellers give shipping price per region, they have to set one by one or aliexpress's default absurdly-high prices win
<jinzo_>
didn't know that - Thanks for the tip!
<oliv3r>
Netherlands!
<mnemoc>
jinzo_: in that case pick austria and add a comment
<jinzo_>
I just tought noone on aliexpress uses China Post Air Mail
<mnemoc>
oliv3r: for .nl, pick .de
<jinzo_>
mnemoc, I can even pick it up in austria (30min by car) if nothing else - again thanks!
<mnemoc>
jinzo_: it's just that sellers usually forget to set custom rules for slovenia
<jinzo_>
can't blame them :D
<mnemoc>
:)
<oliv3r>
I know hipboi's store/stuff is reasonably popular in NL, hipboi, add .nl to your list of china post destinations :D
<oliv3r>
makes things much easier and logical for us stupid europeans :p
<jinzo_>
:D
<oliv3r>
as I said, I don't mind payinga few dollars more, with that sponsoring opensource devs :)
<jinzo_>
or at least put a note somewhere - that will increase the sales in my hones opinion
traeak has quit [Ping timeout: 240 seconds]
<oliv3r>
that too :)
<jinzo_>
s/hones/honest/
<ibot>
jinzo_ meant: or at least put a note somewhere - that will increase the sales in my honest opinion
Omni-Max__ has joined #arm-netbook
<jinzo_>
oo goodie :)
<hipboi>
good idea
<jinzo_>
That means cubieboard to EU is ~40€ - that's awesome. I hope it's sustainable.
RITRedbeard_ has quit [Read error: Connection reset by peer]
<hipboi>
yes, i will have to live on it
<mnemoc>
jinzo_: just don't forget to buy some accessories to let hipboi's family eat :p
<oliv3r>
A note, and slowly add shipping destinations, add once, never worry :)
<hipboi>
yes, F10 contains ir and 2.4G wireless
<hipboi>
but no ir dongle
traeak has joined #arm-netbook
<hipboi>
only a rf dongle
<jinzo_>
that's Great - just tought it's using IR.
<oliv3r>
I will then possibly create an account and order some new toys from your shop hipboi ;)
<oliv3r>
hipboi: btw, your domain is still broken, hipboi.org
<hipboi>
oh, my Chinese blog
<hipboi>
hosted on Dreamhost free host
<oliv3r>
I googled it once and found it to be dead; hoped for a link to your webshop
<hipboi>
they don't provide free host any more
xxiao1 has joined #arm-netbook
cat_n9 has quit [Ping timeout: 248 seconds]
xxiao has quit [Ping timeout: 240 seconds]
<oliv3r>
hipboi: one more question, the PHY you use for the ethernet on A10, does it have the same problems as the one on the a1000? I read somewhere that the a1000 was using pio mode or something for the datastransfer crippling network performance so bad, that a USB dongle would actually work better
* specing
waves to jinzo_
* jinzo_
waves back o/
IEFtm has joined #arm-netbook
IEFtm has quit [Client Quit]
<hipboi>
oliv3r: it's the same phy as mele
<oliv3r>
bummer :( I wonder how bad performance actually is :)
<oliv3r>
i read that blu-ray decoding over the network, or other high-bitrate encodings would chop
<hipboi>
indeed
<oliv3r>
but taht
<oliv3r>
but that's a bug in the a10, not the phy, right?
cat_n9 has joined #arm-netbook
<oliv3r>
you could use an a10 based tablet in your store :p
IEF has quit [Quit: stupid screen.]
IEF has joined #arm-netbook
kaspter has quit [Remote host closed the connection]
<hipboi>
oliv3r: yes
Simon_G has joined #arm-netbook
IEF has quit [Quit: leaving]
IEF has joined #arm-netbook
cat_n9 has quit [Ping timeout: 248 seconds]
ibrah has quit [Quit: ChatZilla 0.9.88.2 [Firefox 15.0/20120827125422]]
<hno>
oliv3r, the Ethernet performance problems is not about the PHY, it's about the MAC that is within A10.
<naguirre>
somebody get opengl acceleration working on a10 ?
<hno>
The driver is far from optimal, but hard to work with as we know very little about the MAC hardware.
<rz2k>
hipboi: you've missed my question or there is nothing new? I'm sorry for being pain in the ass, but, any news about mali r3p1?
<hipboi>
rz2k: news?
* rz2k
trying to understand why our drm and mali.ko dont want to be friends.
<hipboi>
rz2k: really
<rz2k>
hipboi: can you deliver r3p1 libs for us like before?
<hipboi>
let me check
<mnemoc>
and please! armhf cedarx libs
<rz2k>
also, did you test r3p0 on linux? it fails for me with unknown symbol in OpenVG lib, cant tell right now which exactly, my r3p0 setup was on sdcard that died last night.
<mnemoc>
rz2k: using the next_mali kernel branch?
<hipboi>
cedarx libs has some serious memory leak problem
<hipboi>
rz2k: really?
<rz2k>
mnemoc: yes, I've tested both
<hipboi>
but someone has used the r3p0 in developing of their tv products
<rz2k>
its crippled library
<mnemoc>
rz2k: commits for that branch are also welcomed :)
<hipboi>
they need openVG for qt
<rz2k>
strange
<hipboi>
an Indian guy who lives in Taiwan
<hipboi>
not sure if he is here
<mnemoc>
hipboi: any chance to convice your colleague to fix it?
<hipboi>
i think he is fixing it
<hipboi>
not sure about the progress
<hipboi>
he insisted there was no bug
<hipboi>
but finally he realised the bug
<naguirre>
:)
<mnemoc>
classic :)
<hipboi>
: )
mysteryname has quit [Quit: mysteryname]
cat_n9 has joined #arm-netbook
<rz2k>
if that guys are developing for qt using oVG, they use mali with x11 and have working setup, or they can just use openvg and raw fbdev?
<oliv3r>
hno: a see, I knew I read that somewhere. it's really a bummer. I wonder if that'll be fixed in the new and mighty a20! (or whatever it will be called)
<orly_owl>
a15 seems to be the upcoming soc
<orly_owl>
iirc
<mnemoc>
yes, and it seems it will be a quad cortex-a7
<oliv3r>
wasn't the a13 'weaker' then the a10? or 'less complex',
<orly_owl>
yep
<orly_owl>
aimed at tablets
<mnemoc>
and 2/4 layer boards
<orly_owl>
no hdmi iirc
<mnemoc>
a12 has hdmi, but I've never heard about a device using it
<oliv3r>
so calling the next one A15, sounds confsuing :p
<mnemoc>
oliv3r: marketing :)
<oliv3r>
so true
<orly_owl>
there's one called a12?
<RaYmAn>
calling it a15 is just really silly :/ unless it's actually a Cortex A15 design :(
<oliv3r>
so the 'budget' version of the a15 will be A18? :p
<mnemoc>
oliv3r: :p
<rz2k>
I would enjoy 2layer board with DDR
<oliv3r>
hence, A20 makes somewhat sense, then they can use A23 for their budget version
<mnemoc>
RaYmAn: that's exactly the marketing-wise beauty
* rz2k
pretty shure that is impossible
<mnemoc>
RaYmAn: cortex-a7 is the little brother of cortex-a15, while 100% compatible with it
<RaYmAn>
It's still poor marketing :(
<mnemoc>
RaYmAn: calling their cortex-a17 as A15 makets people buy it more
<mnemoc>
a7
<mnemoc>
damn. coffee
<WarheadsSE>
heh
<mnemoc>
RaYmAn: calling their cortex-a7 as A15 makes people buy it more
<RaYmAn>
poor in the sense of, well, cheating
ssalevan has joined #arm-netbook
<mnemoc>
RaYmAn: chinese marketing works that way :|
<mnemoc>
the add the Hz of each unit in the chips too
<RaYmAn>
yeh
<mnemoc>
it's not strictly a lie... just... not accurate
rellla has quit [Quit: rellla]
<jelly-home>
mnemoc: that's the definition of "lie"
<jelly-home>
just sayin
<popolon>
sounds like some are confusing between arm cotex A8,A9, A15, A7 and Allwinner A10, A13
<RaYmAn>
jelly-home: it's not a lie - more like twisting the truth :P
<jelly-home>
RaYmAn: stop proving my point
<oliv3r>
they are 'abusing' the similarity of names to make it seem it is more then it actually is
<oliv3r>
i think that's as lcose to lieing as you can get, without actually calling it lieing
<RaYmAn>
you can't really call the naming lying as such. They aren't calling it "Cortex A15", they are calling it "Allwinner A15" etc
<RaYmAn>
A15 by itself means nothing, despite people using it.
<oliv3r>
yes, they make it appear as if it where a cortex a15, without actually saying it
<oliv3r>
it's misleading
<oliv3r>
which i think is almost, if not, lieing
<RaYmAn>
I suspect there is a clear legal distinction :P
<popolon>
and there are some tablets called A10
<oliv3r>
clear, or chinese :p
<popolon>
that doesn't use allwinner chips
<oliv3r>
i would think a15 to be less confusing as the whole a10/a13 thing, especially if they choose to call the butcherd version a14 :p
<RaYmAn>
If one could argue that it was common knowledge that A15 == Cortex A15, then maybe. Otherwise it's just people not understanding A15 != Cortex A15. There's no doubt it's misleading marketing, but lying; not so much.
<oliv3r>
almost lying :p
<oliv3r>
shenzen-style :p
<oliv3r>
jk
<popolon>
they are not from shenzhen
<oliv3r>
gangdam style would have made less sense, as that's korean :p
<orly_owl>
lol
<popolon>
like windows didn't discovered real windows nor system computer windows
<popolon>
americans are all big liers too ;)
<popolon>
that's a little stupid to associate action from some people to a whole country
<oliv3r>
no denying that :p
<popolon>
A10 for Allwinner 10 could be understandable
<mnemoc>
marketing is by definition the art of making you believe you need/want something you don't really do
<popolon>
intel call it's last processors core i3, core i5, core i7 isn't it ?
<popolon>
rockchip rk3066...
<mnemoc>
as also happens to lawyers and politians, marketing people can't really do their job well without adding some... not strictly true salt
<hno>
jelly-home, it's not a lie, it's lawyer & marketing default style.
<jelly-home>
hno: marketing is the ART of lying
<oliv3r>
if allwinner made intel's chip, it would have been i3, i7, i5 (in order of performance)
<oliv3r>
anyway, it's all fun and games and once 'we' figure out which chip does what
<oliv3r>
the name means only so much
sspiff has joined #arm-netbook
sspiff has quit [Changing host]
sspiff has joined #arm-netbook
<oliv3r>
I wonder if the next-next iPhone will use allwinner's A10, it beats apple's own A6 :D
<oliv3r>
in naming :p
<jlj>
Is it possible to build the sndi2s,sun4i_i2s
<popolon>
apple a5 is a samsung implementation of cortex A9 :D
<jlj>
drivers as modules? I seem to be getting messages like "sun4i_sndi2s cannot find any using configuration for controllers, return directly" when I try
<rz2k>
mali_drm from r4p1 pack has driver.platform_device = dev; in that place, seems like guy who pushed it to git just deleted that line because in 3.x it is now driver.kdriver.platform_device.
* rz2k
ping techn_
<hno>
hi
<rz2k>
s/r4p1/r2p4/
<ibot>
rz2k meant: mali_drm from r2p4 pack has driver.platform_device = dev; in that place, seems like guy who pushed it to git just deleted that line because in 3.x it is now driver.kdriver.platform_device.
<rz2k>
hi hno
<naguirre>
mnemoc: there is a r3p1-01rel0 version on the arm website, did you tried it ?
pawel5870 has joined #arm-netbook
<mnemoc>
naguirre: nope. we have r3p0 on the next_mali branch (tested by Turl)... patches to update that to r3p1 are welcomed... but we are limited by the availability of r3p1 libs
<naguirre>
rz2k: next_mali branch do not include this driver ?
<mnemoc>
rz2k: can't :< awfully busy :|
<rz2k>
naguirre: next_mali is just r3p0 from malideveloper.com, mali_drm is still from r2p4
<mnemoc>
fix it :)
<rz2k>
mnemoc: ok, I will post that there.
<naguirre>
and the changes are interresting ?
<naguirre>
between r2p4 and r3p0
<naguirre>
i can't find a changelog
Simon_G has quit [Ping timeout: 246 seconds]
L84Supper has quit [Quit: puff of smoke]
<hno>
mali_drm is from Tom. It's not part of the ARM MALI driver but SoC platform specific.
pawel5870 has joined #arm-netbook
<hno>
except the one we have is almost an empty shell.
L84Supper has joined #arm-netbook
cat_n9 has joined #arm-netbook
jinzo_ has joined #arm-netbook
cat_n9 has quit [Ping timeout: 264 seconds]
<cat_x301>
rz2k: there is no any other mali_drm afaik
<cat_x301>
oops.. hno, haven't noticed your comment
<cat_x301>
rz2k: anyway, i understood that mali_drm is rather generic and was only supplied once with r2p4..
repvik has quit [Remote host closed the connection]
<mnemoc>
ours is actually no more than an skeleton...
von_fritz has quit [Quit: vonfritz leaves, don't panic]
cat_n9 has joined #arm-netbook
<hno>
skeleton or empty shell, mostly the same just different animals.
<hno>
inner or outer skeleton.
tzafrir has quit [Ping timeout: 268 seconds]
<mnemoc>
:)
<mnemoc>
hno: indeed your "empty shell" is more correct as they are NOP implementations of the needed hooks
jinzo_ has quit [Ping timeout: 246 seconds]
popolon has joined #arm-netbook
Quarx has quit []
<popolon>
do you know if there is an arm board with HDMI input for using VPU/dsp to encode HDMI signal and push it on the network ?
<techn_>
what I remember r3p0 didn't have mali_drm anymore on ARM's site.. previous one's did
<techn_>
rz2k:
<rz2k>
yep
<techn_>
+ mali_drm in our repo is broken.. open/probe logic won't work -> no drm in our dev-tree
<techn_>
I tried with linaro's one and I managed to get drm to dev-tree.. but still no opengles2 :(
<techn_>
could we use opengl libraries without X? with directfb? or wayland?
<techn_>
btw. which rendering library android is using?
<rz2k>
do you have logs with mali init when drm is in dev-tree?
<cat_x301>
popolon: +1
<techn_>
not currently..
ZaEarl has quit [Remote host closed the connection]
jlj has quit [Ping timeout: 245 seconds]
<popolon>
Most SoC with Cortex A8/A9 has a DSP or vpu able to encode 1080p in h264 today (with good quality, when I look at what do my phone). but the optic & the sensor are only good in day light
<popolon>
I would like to plug my camera HDMI output to these good encoder directly :p but don't found the good solution :/
<techn_>
rz2k: yep.. http://git.linaro.org/gitweb?p=people/chunsangjeong/mali-dev.git;a=blob;f=drivers/gpu/drm/mali/mali_drv.c or is meego's one.. but both looks better than ours
<traeak>
get a usb device...but the usb device of course would have an encode chip on it
<traeak>
so it woudl take over the hard work
<rz2k>
mali400 is on PCI? because there is a mention of it in function names
<rz2k>
return drm_pci_init(&driver, &mali_pci_driver); for example
eFfeM has joined #arm-netbook
<techn_>
there's other http://www.igloocommunity.org/gitweb/?p=android/platform/vendor/st-ericsson/multimedia/linux/mali400ko.git;a=blob;f=x11/mali_drm/mali/mali_drv.c;h=30538f5932e753296767821d0c088d88ed6c9ca9;hb=HEAD
jinzo_ has joined #arm-netbook
rsalveti has quit [Ping timeout: 260 seconds]
<techn_>
rz2k: .. it could be this row that we are missing "driver.kdriver.platform_device = dev;" .. like you earlier mentioned
<rz2k>
also, why igloo's mali_drm has two device initializations?
jlj has joined #arm-netbook
eebrah_ has joined #arm-netbook
<techn_>
for both fb'?
eebrah_ has quit [Ping timeout: 240 seconds]
tzafrir has joined #arm-netbook
eebrah_ has joined #arm-netbook
<naguirre>
whe you say empty shell what's is missing ? I don't understand what is missing to have a complete chain working.