<libv>
and then for europe you have to take customs into account
<libv>
putting it at 500EUR
<mnemoc>
the chromebook is better idea :p
<libv>
definitely
<mnemoc>
mine is already in the town of the coworker who will forward it from usa for me :)
<libv>
yeah, i read that from the backlog
<libv>
lucky you :)
<mnemoc>
happy happy
<mnemoc>
the dumb part is that I don't need it at all :<
<mnemoc>
but well... it's a toy. one doesn't need to need the toys to want them or play with them
shivansps has joined #arm-netbook
<mnemoc>
anxious anyway
<mnemoc>
early xmas present
<shivansps>
im wondering, evtouch for xorg should work with A10 touch screens?
<mnemoc>
should....
<shivansps>
i dont have experience playing with x11, whiout a xorg.conf, where i should point out the evtouch driver? i can add a xorg.conf only for the touch input? i
<mnemoc>
it should be magic.... if the kernel detected your TS
<shivansps>
i bet that is in the script.bin...
<mnemoc>
the settings, yes
<mnemoc>
but you need to have the proper driver in your kernel anyway
<shivansps>
IM using the script.bin of the tablet, LCD and Wifi works, somehow sound and touch do not, as well as webcam.
<shivansps>
the touch module its there and well its the sound ones
<shivansps>
odd
<shivansps>
because the same script.bin works for the android kernel
<mnemoc>
we don't have all the drivers included in the gpl violating kernels distributed with the devices
<shivansps>
but sounds works with other script.bin, older ones anyway
<mnemoc>
diff them
<shivansps>
as for touch, its the goodix_touch, im 99% sure but ill double check
<mnemoc>
and take the parts relevant to sound into the one that works most
<shivansps>
ill boot intro android and double check the ts driver
<mnemoc>
capture the dmesg so you can compare
slash_random has joined #arm-netbook
<drachensun>
shivansps: What tablet? my touch screen is working and the camera comes up
<drachensun>
but there is a bug with the camera when called by a capture program
<drachensun>
kernel crashes, I haven't gotten back to it yet
<shivansps>
mmm and i guess i need to compile in blacklight support in order for the kernel to shut down lcd backlight when it shut downs the screen for inactivity
<shivansps>
novo 7 aurora
<drachensun>
oh ok, not the same as mine, its got the ft5x touch screen
<drachensun>
mine does I mean
<drachensun>
I'm trying to figure out the hardware buttons too
<rz2k>
aurora has spi touchscreen
<rz2k>
you need a separate patch for that, afaik.
<rz2k>
dont know if was merged or not
<rz2k>
techn_: I've checked right now, I have same xbmc crash as you have on r3p0 and working xbmc on r3p1 :/
<drachensun>
I traced them with a meter and I think they run to the AXP209 so I thought I should be able to see them with the AX2 gpio driver but no luck
<mnemoc>
lcd_spi support is merged
<mnemoc>
but relies in script.bin
<shivansps>
well the goodix ts is on the dmesg on linux
<mnemoc>
same output?
<shivansps>
give me a sec...
<shivansps>
man i keep writing "ls" in windows -.-
<shivansps>
as a side note, the goodix only apears once i modprobe goodix_touch, it does not apear alone
<shivansps>
Nice A10 netbook, but WTH they keep evading adding the sata????!! its that difficult to use a few more pins of the A10 chip?
popolon has quit [Quit: Quitte]
<Gumboot>
WarheadsSE: I remember asking you about the SATA interface on nit6x, and at the time there was a reason you couldn't answer, but I forget what that reason was. Did that reason go away?
stefanro has joined #arm-netbook
stefanro1 has quit [Ping timeout: 276 seconds]
<drachensun>
so I keep getting this EGL errors with xbmc, do I need the mali libs installed maybe?
<mnemoc>
likely :)
<mnemoc>
good night
<drachensun>
heh I feel kinda dump once I unzipped it
<drachensun>
dumb
<drachensun>
saw the names
<drachensun>
ok thanks
<shivansps>
someone has managed to compile it on armhf???? i can only do it on armel, on armhf complains about missing SQLITE3 and its installed....
<drachensun>
rellla and empat0 both were reporting it working earlier
<shivansps>
i dont think they are compiling it natively
<shivansps>
anyway, ill crosscompile it, it seems the only way
<rz2k>
check maybe sqlite3 is named wrong
<rz2k>
for example when compiling libump with dri2 you need to symlink libdri2.so.1 to libdri2.so because it is not in ld.conf.d
valhalla has quit [Ping timeout: 255 seconds]
tuliom has quit [Quit: Konversation terminated!]
L84Supper has quit [Quit: puff of smoke]
<drachensun>
I managed to cross compile it today
<drachensun>
but I'm getting EGL errors trying to run it
L84Supper has joined #arm-netbook
<drachensun>
following relllas instructions on the wiki pretty much worked but I used empat0's repository
<shivansps>
sqlite3 3.7.9-2ubuntu1 Command line interface for SQLite 3 libsqlite3-dev 3.7.9-2ubuntu1 SQLite 3 development files
<shivansps>
drachensun: you have mali driver loaded? its showing up on lsmod?
<shivansps>
checking for SQLITE3... no configure: error: Could not find a required library. Please see the README for your platform. FUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUU
slash_random has quit [Ping timeout: 240 seconds]
<shivansps>
on armel this does not happen
<drachensun>
yeah, the mali driver is loaded
<drachensun>
about the sqlite thing
<drachensun>
I had to make a link from arm-linux-gnueabi to (the real) arm-linux-gnueabihf
<drachensun>
something kept trying to force a check to the wrong location
<drachensun>
maybe that is your sql problem?
<drachensun>
that came up during my build of the dependencies I think
<WarheadsSE>
Gumboot: I had a problem with sata because of the soldering job of the gen1 I had. I have recevied a replacement that has working sata.
<shivansps>
OK i downloaded the sqlite3 source,compiled it and installed, its going OK now... now its complaning about php... ill have to do this for EVERYTHING?
mSquare has joined #arm-netbook
<shivansps>
npg i men
<shivansps>
png
<shivansps>
i must compile libpng too it seems, wth is this broken packages?
<rz2k>
are you sure you did apt-get build-dep xbmc ?
<rz2k>
this will install stuff that was used to compile xbmc (according to package headers)
<shivansps>
yes i did it, packages are installed
<shivansps>
and it works on armel i can confirm you that
<drachensun>
weird
<shivansps>
armhf package are installed but not detected
<rz2k>
interesting
<drachensun>
I didn't have any of that
<drachensun>
what rootfs do you have? linaro 12.09?
<shivansps>
linaro 12.08
<rz2k>
as a guess: try go apt-get build-dep xbmc-bin
<shivansps>
im using the one from j1nx
<rz2k>
how I remember xbmc is two or three packages
Svet has joined #arm-netbook
Svet has joined #arm-netbook
Svet has quit [Changing host]
<drachensun>
rz2k do you know how all these mali builds relate?
<shivansps>
xbmc-bin redirects to xbmc
<drachensun>
There are downloaded binaries libraries and then the build instructions, are they both needed? do they both have to be matching r3p0 or r3p1?
discopig has quit [Ping timeout: 276 seconds]
<Gumboot>
WarheadsSE: How's the SATA interface attached? Is it a first-class peripheral, or is it on USB or something horrible like that?
<WarheadsSE>
its a PHY attached 1st class
Svet is now known as sv
<Gumboot>
Oh. Well, then. It sounds like my dream hardware.
<Gumboot>
What speed ethernet?
<WarheadsSE>
We have 1 in the Arch Linux ARM build farm
<WarheadsSE>
it beats the piss out of the Pandas
<WarheadsSE>
GbE PHY
<WarheadsSE>
I'll have to see what I can get through it later
<WarheadsSE>
It's a bit late and I have a Conf. tomorrow
<Gumboot>
It's about the same price as a panda, isn't it?
<WarheadsSE>
it will be $199 w/o wifi in november
<WarheadsSE>
pandas run about $180-190 USD
<Gumboot>
So they haven't seen mass production, yet. This is why you got a dodgy solder job, is it?
<Gumboot>
It's not proprietary information or anything, right? I can just re-post it into pastebin with a longer life?
<WarheadsSE>
This is one sent to me by BD to get ALARM running on
<WarheadsSE>
And AFAIK, there was no NDA involved
<Gumboot>
OK, coo.
<Gumboot>
cool.
<Gumboot>
Thanks.
<WarheadsSE>
You can see at the end where I tried to mount a raw device :p whoops
<rz2k>
drachensun: you mean rXpX versions?
<shivansps>
libpng compiled and installed, and now checking for PCRECPP... no configure: error: Could not find a required library. Please see the README for your platform.
<shivansps>
this is some kind of joke
sv has quit [Read error: Connection reset by peer]
tekzilla has quit [Ping timeout: 255 seconds]
tekzilla has joined #arm-netbook
sv has joined #arm-netbook
L84Supper has quit [Ping timeout: 256 seconds]
<drachensun>
rz2k: yeah
<shivansps>
checking for SAMPLERATE... no configure: error: Could not find a required library. Please see the README for your platform. OK whats that?
L84Supper has joined #arm-netbook
avernos has quit [Remote host closed the connection]
<drachensun>
did you build the dependencies without problems?
<drachensun>
under tools/a10/depends
<shivansps>
so far, sqlite3, png and pcre libs
<shivansps>
i had to downloads the sources and compile them
<shivansps>
i build xbmca109 before in armel, i known how to do it
<drachensun>
well something is wrong
<drachensun>
I mean did you do this step "/melehacking/xbmca10/tools/a10/depends# make" and have no problems?
<shivansps>
yeah that and the liblame too, ok
<shivansps>
configure its complaning about depedency libs that build-dep xbmc should take care about
<shivansps>
the libs are installed but does not work
<drachensun>
because I had to create that library link there to find some dependency
<shivansps>
that should not happen that whats configure about
<drachensun>
ok
<drachensun>
well good luck
<WarheadsSE>
well, this could make building XBMC for the A10 cake on arch..
<shivansps>
it seems like broken packages to me
<shivansps>
the thing is far too many
<drachensun>
I'm telling you man, I think xbmc is hardcoded to look in /usr/lib/arm-linux-gnueabi
<drachensun>
and on an hf system its all in /usr/lib/arm-linux-gnueabihf
<WarheadsSE>
so symlink :p
<drachensun>
yeah thats what I did
<WarheadsSE>
the glibc update changed a lot of things.
<shivansps>
i dont have a arm-linux-gnueabu folder, just HF, why is finding GCC and the other libs them? this is not the case
tzafrir_laptop has quit [Ping timeout: 246 seconds]
t0dbld has joined #arm-netbook
avernos has joined #arm-netbook
avernos has quit [Changing host]
avernos has joined #arm-netbook
pwhalen has quit [Ping timeout: 252 seconds]
merbanan has quit [Ping timeout: 252 seconds]
pwhalen has joined #arm-netbook
Quarx has joined #arm-netbook
tzafrir_laptop has joined #arm-netbook
<shivansps>
ok its compiling now, i have to download and compile sqlite3, png, pcrec, samplerate and freetype2
<shivansps>
it still complaning about some not essential libs that are installed, like libusb... even ALSA... maybe thats why sound its not working for me
<shivansps>
strange
mysteryname has joined #arm-netbook
mysteryname has quit [Remote host closed the connection]
drachensun has quit [Quit: Leaving]
alcides` has joined #arm-netbook
alcides has quit [Ping timeout: 244 seconds]
alcides` is now known as alcides
alcides has quit [Changing host]
alcides has joined #arm-netbook
<orly_owl>
which mk802 has an hdmi plug? mk802 II?
RITRedbeard has quit [Read error: Connection reset by peer]
<rz2k>
he claims that mnemoc, hno and hipboi made A10 hacking die, srsly, wtf?
alcides` has joined #arm-netbook
alcides has quit [Ping timeout: 244 seconds]
tmerle has quit [Ping timeout: 245 seconds]
avernos has quit [Ping timeout: 246 seconds]
avernos has joined #arm-netbook
avernos has quit [Read error: Connection reset by peer]
avernos has joined #arm-netbook
IEF has quit [Ping timeout: 252 seconds]
arnd has joined #arm-netbook
IEF has joined #arm-netbook
tmerle has joined #arm-netbook
tmerle has quit [Ping timeout: 252 seconds]
tmerle has joined #arm-netbook
tmerle has left #arm-netbook [#arm-netbook]
Undertasker1 has joined #arm-netbook
zub has left #arm-netbook ["mv MK802 garbage"]
<mnemoc>
rz2k: how?
eFfeM has joined #arm-netbook
<mnemoc>
rz2k: if any onf the changes in our u-boot or linux trees don't please them, they can always revert or fork... that's what git is best at
<RaYmAn>
they seem mostly to just be whiners tbh
popolon has joined #arm-netbook
<RaYmAn>
also, he's complaining about nand support being removed...in february? lol
<mnemoc>
the usage of "Amery" is kind of disturbing.... he used it like it was a one word name...
<mnemoc>
very weird that gnexus guy is
<Maqs>
:D
<techn_>
rz2k: damn.. we should try move to r3p1 then.. only android libs missing?
<rz2k>
techn_: yeah
<rz2k>
://
<techn_>
mnemoc: backup is probaply there becouse if you install mesa-egl, everything is overwritten :/
<techn_>
mnemoc: maybe better solution would be just have bins on /lib/rXpX and symlink to /lib
<techn_>
or make dep file which contains those libs and plays as egl provider
<mnemoc>
techn_: any comments about the changes I did tonight to the mk_hwpack script?
<techn_>
havent checked them yet
<techn_>
lol@doozan
oliv3r has quit [Ping timeout: 245 seconds]
ol1ver has quit [Ping timeout: 246 seconds]
<mnemoc>
I stopped reading at the part he claims that the cubie is a beagleboard/pandaboard knockoff
<RaYmAn>
isn't it in some ways? But only as far as "cheap dev board" goes =P
<RaYmAn>
(beagleboard was insanely cheap comparatively when it came out)
<mnemoc>
like all phones with rounded corners are iphone clones :p
<RaYmAn>
exactly :P
<RaYmAn>
He's clearly just a kid that got annoyed with things not progressing like he wants and rage-quit
<mnemoc>
and loves his galaxy nexus
<mnemoc>
(I love mine too)
tzafrir_laptop has quit [Ping timeout: 256 seconds]
rellla has joined #arm-netbook
<RITRedbeard>
Q: is the sata on Cubieboard/A10 near full speed?
<RaYmAn>
I'm waiting for next nexus phone, heh =P
tuliom has joined #arm-netbook
<mnemoc>
RITRedbeard: I get 60MB/s with a normal 2.5 hdd powered by the board, and 120MB/s when using an externally powered 2x3.5" raid1 device over esata
<RITRedbeard>
is that expected?
<mnemoc>
not real "full speed" but pretty damn good for this cpu imho
tuliom has quit [Client Quit]
rellla has quit [Ping timeout: 255 seconds]
rellla1 has joined #arm-netbook
tzafrir_laptop has joined #arm-netbook
<techn_>
mnemoc: yes.. it cleaner now
<mnemoc>
techn_: makes sense?
<mnemoc>
but I think mk_hwpack should receive 2 arguments, 1) abi (android, armel, armhf), 2) output file
<mnemoc>
the rest can easily be extracted from the config file
<mnemoc>
or maybe we add the abi and hwpack format to the config file....
<mnemoc>
my build server still runs ubuntu 10.04... I don't have hf toolchain or 7z there :|
<techn_>
not sure yet.. I'll need leave to grosery shop
<techn_>
*grocery
<mnemoc>
:)
<mnemoc>
btw, I tried to separate the interim constructs in build/ and leave only the final files at output/
<mnemoc>
still unsure if $board.bin is interim or final :|
<techn_>
yep.. That was my tought too
<mnemoc>
both can be different on different partitions, build/ could even be tmpfs and output/ remote ... so they need clear boundaries
<mnemoc>
s/different//
<techn_>
I think clean must be improved so that it cleans build .. and not output? or both?
<mnemoc>
i would say build/ only
<mnemoc>
grocery store?
<techn_>
waiting my gf :)
<mnemoc>
:)
* mnemoc
wonders if cnxsoft would hand over access to the monster server he got donated. doing my sunxi builds in the $work$ server isn't very correct
<mnemoc>
and finally get the nightly builds fixed
<mnemoc>
techn_: I also think we should rm the _hwpack dir before (re)creating it
<mnemoc>
Turl: got a harddrive yet :p your monster idling desktop wants to build android images :p
<Maqs>
empat0: what does "Out of decoder frame buffers. Freeing the queue." tell me? is there a way to have more decoder frame buffers?
<empat0>
Maqs: this just means that xbmc was a little slow, e.g. loading menus or network delay. it is not critical, but you might get some artefacts on the screen. the number of decoder buffers is managed within libvecore, no way to change it.
ZaEarl has quit [Ping timeout: 246 seconds]
popolon has joined #arm-netbook
mSquare has quit [Ping timeout: 245 seconds]
<Maqs>
ok, thanks
<Maqs>
and "ERROR: A10: DISP_CMD_VIDEO_SET_FB failed."?
<Maqs>
empat0: it plays videos, but the screen always goes to one color (mostly something white or magenta), I see blocks of pixels and suddenly everything looks okay again.. every few seconds
<Maqs>
seeking crashes after "ERROR: A10: decode(73121): 5"
<empat0>
Maqs: that is critical. it means that a ioctl failed. . you will not be able to use hardware rendering. are you using armel or armhf?
<empat0>
Maqs: i have also observed the seeking crash. im working on this.
<Maqs>
armhf
<Maqs>
recent stage/sunxi-3.0
<empat0>
Maqs: i have done my testing on wip/next_mail,, only short testing on stage kernel. i will look at this.
<Maqs>
does DISP_CMD_... need more than the disp kernel module?
mSquare has joined #arm-netbook
<empat0>
Maqs: afaik no, but perhaps some of the kernel devs can comment.
<empat0>
Maqs: do you see something in dmesg about the ioctl?
<Maqs>
i have some "[DISP WRN] file:drivers/video/sunxi/disp/de_bsp_sun4i/de/disp_layer.c,line:239: not supported image0 pixel sequence:218 in img_sw_para_to_reg" lines, but that's it
<Maqs>
besides that, my xbmc.log is filled with alsa buffer underruns and discontinuity debug lines by CDVDPlayerAudio
<empat0>
Maqs: perhaps you can give me a short sample of a file you have problems with.
alcides` is now known as alcides
alcides has quit [Changing host]
alcides has joined #arm-netbook
<Maqs>
the problem with usb may be that it's too slow
<Maqs>
hdparm -t gives me 12MB/s
<empat0>
Maqs: that may be the reason. i do most of my testing from fast sd card.
mSquare has quit [Quit: Leaving.]
L84Supper has quit [Quit: puff of smoke]
tmerle has joined #arm-netbook
rellla has joined #arm-netbook
gimli has joined #arm-netbook
<merbzt>
empat0: did you get mpeg4 codecs working ?
L84Supper has joined #arm-netbook
andrew__ has quit [Ping timeout: 245 seconds]
rsalveti has quit [Ping timeout: 260 seconds]
<empat0>
merbzt: partial. xvid more or less, vp6, vp8, wmv1, divx3. it was my missunderstanding of ffmpeg codecid.
<merbzt>
nice
mysteryname has quit [Ping timeout: 248 seconds]
popolon has quit [Quit: Quitte]
rsalveti has joined #arm-netbook
<techn_>
uaaah.. we should fix mali Makefiles
<mnemoc>
techn_: rz2k has patched repos for mali libs
<techn_>
kernel modules?
<techn_>
since I have problem with j8.. sometimes compilation fails :(
tmerle has quit [Read error: Connection reset by peer]
<WarheadsSE>
lol
<mnemoc>
-j16 works for me (tm)
<mnemoc>
but sure, those makefiles are junk
<WarheadsSE>
People doing complete cross
<WarheadsSE>
pfft.
<mnemoc>
techn_: BUT, no point in fixing sunxi-3.0's mali considering there is a pending upgrade in the stage branch
tmerle has joined #arm-netbook
<techn_>
mnemoc: same with stage :(
merbzt has quit [Ping timeout: 260 seconds]
eFfeM has quit [Quit: Leaving.]
slash_random has joined #arm-netbook
<libv>
mnemoc: i was full of it.
<libv>
_very_ full of it.
<lundman>
you guys just throw -j16 on there thinking it always makes it faster?
<libv>
value &= bit & bit & bit; actually was value &= mask & mask & mask; with mask each time being ~bit
<libv>
it's one of those things that you can do, but not really should do as it requires way more mental arithmetic than it is worth
<mnemoc>
indeed
<libv>
i now have a commit which tosses these ~bit defines out and properly does &= ~(bits as used in the previous function);
* jelly-home
has a feeling of deja-vu
<mnemoc>
libv: should I pull them to stage yet?
<libv>
mnemoc: patience :p
<mnemoc>
:)
<libv>
this is where i gave in last morning and went to sleep, and i had a good... days? sleep since my cold is starting to ease :)
<libv>
also, i removed braces around a single line statement... and then the whole thing refused to build...
<libv>
there is very little do { } while(0) in defines
<libv>
but most of the defines are something with volatile anyway, this while they should be __iomem
<libv>
there is so much to do here :)
<mnemoc>
no doubt, but without sync'ing the big cleanup works often it starts to conflict fast with the work others might be doing. cat_x301 and "Leith Bade" in the case of disp
<libv>
give me a few more h :)
<mnemoc>
no problem :)
<libv>
i am working on redoing what i had, once i am back to that point, it's all yours :)
<libv>
now i am back to make C=1 :)
<libv>
i will do a diff against /dev/null as well, and checkpatch that, so that all the pain is over with in one go, so that might take me a few h
<libv>
the make C=1 things are pretty fast
<libv>
be warned though, i do not believe in the 80 character limit for data tables :)
<techn_>
mnemoc: cat_x301 told that he has done nothing
<libv>
and two years on, the same people who were religiously sticking to 80 chars moved to 100 chars and some even praised me for it (which is rare, i never get that in the xorg community :))
<libv>
s/on/later/
<ibot>
libv meant: and two years later, the same people who were religiously sticking to 80 chars moved to 100 chars and some even praised me for it (which is rare, i never get that in the xorg community :))
<mnemoc>
:)
<libv>
some of the tables in disp need to be sanitized though, those fir tables need to be segmented, that will cut down logic and code complexity quite a bit
<mnemoc>
when I look at those tables I just panic
<libv>
it's not too bad
<libv>
they are not bare register values like i had to deal with on via
<mnemoc>
it probably because I don't know where the numbers come from
<libv>
i'd guess testing
<mnemoc>
:|
<mnemoc>
totally unmaintainable
<libv>
well, they are filter settings
<mnemoc>
techn_: btw, I pushed a minor change to the bsp's Makefile, but likely to conflict if you have something modified there
<libv>
techn_: what are the license statements on the iep files btw
<libv>
one even has an "all rights reserved" on it
<mnemoc>
the whole tree was initially like that
<libv>
right, back in may or june or so
<mnemoc>
then allwinner authorized hipboi (which he was still working there) to fix them as they were just a bug
<mnemoc>
this iep stuff comes from the SDK
<libv>
ok, does that mean that this applies to these new files as well?
<mnemoc>
yes, they have said that all the kernel code in the SDK is gpl too
<libv>
okies, will fix.
rellla has quit [Quit: rellla]
slash_random has quit [Ping timeout: 240 seconds]
rellla has joined #arm-netbook
rellla has quit [Client Quit]
oliv3r has joined #arm-netbook
<libv>
hrm... my feeling is that disp no longer builds without UMP enabled
<mnemoc>
disp_ump was supposed to do the glueing..
<libv>
but the way disp_ump stuff was thrown into dev_disp.h massively broke separation and now the whole thing seems pretty grey
<libv>
trying to fix up the header fallout now
<libv>
lcd/hdmi modules were defining disp symbols in local headers :(
<mnemoc>
:<
<libv>
so those symbols were made static in disp, which then ran into undefined symbols...
<libv>
my first pass at this was rather... simplistic
<libv>
now i am doing it properly, and noticing a lot more issues :)
<libv>
i do not get why disp_ump needs to be another module.
<libv>
the way the logic for dealing with one of its symbols is handled is unnecessarily complicating matters imho, but i haven't looked deep enough
<libv>
hrm, surely there is a better way, but later.
mikey_w has quit [Remote host closed the connection]
<libv>
hrmpf. how is the module dependency tree really built up here.
<libv>
ok, i think i found the right separation now for this header. dev_fb.h
vinifm has joined #arm-netbook
ol1ver has joined #arm-netbook
tuliom has joined #arm-netbook
mikey_w has joined #arm-netbook
n6pfk has joined #arm-netbook
TheLarch has joined #arm-netbook
<RaYmAn>
semi-offtopic - Would it be possible to bitbang serial output on an ARM board? from within linux.
L84Supper has quit [Ping timeout: 245 seconds]
TheLarch is now known as L84Supper
<L84Supper>
RaYmAn, if the hardware is there (e.g. GPIO) sure
<mnemoc>
RaYmAn: if in A10, the current gpio interface is very inefficient, but theorically you can refactor the code of sunxi-tools' `pio.c` into a lib and try on /dev/mem directly
eFfeM has joined #arm-netbook
<RaYmAn>
I'm thinking in-kernel driver type thing
<RaYmAn>
(not a10)
<mnemoc>
with gpiolib i suppose
<RaYmAn>
I'm mostly wondering if it's fast enough
<RaYmAn>
or accurate enough
<mnemoc>
iirc tsvetan measured ~400MHz bitbanging on gpio from userspace on his a13 olinuxino
<mnemoc>
s/ on / one /
<ibot>
mnemoc meant: iirc tsvetan measured ~400MHz bitbanging one gpio from userspace on his a13 olinuxino
gzamboni has quit [Remote host closed the connection]
<shivansps>
mnemoc: any idea of what could be causing the goodix ts to do not work, this is the dmesg of android and linux http://pastebin.com/mWNVhAfU
gzamboni has joined #arm-netbook
<mnemoc>
both look identical to me
<shivansps>
the linux one has a few extra lines and missing the goodix_init_panel *res*
<mnemoc>
RaYmAn: going to linaro connect?
<RaYmAn>
mnemoc: was considering it, but it's during my work hours and stuff
<RaYmAn>
so free quickly becomes expensive :P
<mnemoc>
shivansps: i have no knowledge about i2c/ts, hand around for a while and someone might give you a useful hint or something
<mnemoc>
RaYmAn: :<
<RaYmAn>
It's not really close enough to my work area to convince them to send me there ;)
<shivansps>
and here it comes Windows RT to complicate things on ARM front even more -.-
<mnemoc>
hopeful MS's idea of making their own hardware makes some large manufacturers divorce from them
<libv>
that netbook doesn't look half nice
<libv>
it will be a dog with openoffice though, which is why i do not see my gf or mother using it
<mnemoc>
docs.google.com :p
<RaYmAn>
these chromebooks are actually tempting for $family
<RaYmAn>
less support :P
<shivansps>
i dont think so, ARM is now capable of invading netbook and notebook markets performance wise, but it hust cant sell a notebook with android and hope people take it seriusly, now that will change a bit with Windows RT... i dnt think i i dont hope that Windows 8 and RT will be succesfull, but they will have a nice market right there
<shivansps>
thats the thing with windows, it sells
<RaYmAn>
until people realize none of their x86 apps work
<mnemoc>
i find so impressive they didn't make it a pure .net/CLR OS...
<RaYmAn>
they did
<RaYmAn>
research-wise
<mnemoc>
iirc they invented a whole new incompatible crap
<shivansps>
thats a stupid move of ms, they could esily integrate a x86 translator intro the os
<RaYmAn>
"easily" - right
<shivansps>
they rater sell you new apps... well its not so stupid move after all
<mnemoc>
that what the .net vm was supposed to do.... like java's
<RaYmAn>
yeah, true - .net apps should actually work
<RaYmAn>
pure .net that is
popolon has joined #arm-netbook
<shivansps>
why now? android can translate, arm apps runs on mips and 86 processors, why MS can not?
<RaYmAn>
There's a long way from using a high-powered x86 to emulate an old or slow arm and using an arm to emulate a high powered x86 :P
<RaYmAn>
It's of course *possible*
<RaYmAn>
The question is whether it runs at acceptable performance.
<shivansps>
the A10 can emulate a x86 with QEMU, and run WinXP almost usable, hell i even try to run some old games on it, and the A10 is everything except a fast cpu... im sure x86 binary can be translated to arm at at aceptable speed, on Android can be done already, and the performance hit is small, check how the Atom Z2460 can run ARM apps, or how JZ4770 can run arm apps too... i have a jz4770 the performance its ok
<penguin42>
shivansps: What makes you say android can do ARM
<penguin42>
sorry
<penguin42>
shivansps: What makes you say android can do ARM translation onto x86?
<RaYmAn>
apps are made for low-powered hardware, that's certainly a lot easier :)
<RaYmAn>
Also, windows xp - lets see:
<RaYmAn>
Pentium 233-megahertz (MHz) processor or faster (300 MHz is recommended)
<RaYmAn>
At least 64 megabytes (MB) of RAM (128 MB is recommended)
<RaYmAn>
So you just need to emulate that to get it running vaguely ok.
<shivansps>
its not android, its an external app that intercepts ARM calls and translate them in real time
vinifm has quit [Quit: Ex-Chat]
<shivansps>
on the JZ4770 mips cpu, there is a android app that let you run ARM apps normally
<RaYmAn>
and it works on eveyrthing? :P
<shivansps>
Intel includes the same thing on the android for thier Atom Z2460 x86
<RaYmAn>
Let me know after you've tried running current x86 desktop apps on an arm device, emulated.
<popolon>
atom are slower than today arm with cortex A9
slash_random has joined #arm-netbook
<popolon>
at least on most benchs
<popolon>
and only for cpu part
<RaYmAn>
I'd imagine running android apps would be done by emulating the native parts, then just running a ported dalvik to run the java parts
<RaYmAn>
incidentally, the jz4770 one only works on a limited set of apps, doesn't it?
<shivansps>
well i did run Caesar 3 on a LiveXP in QEMU, it was like 2 fps but thats a worst case escenario, live xp makes things worse, and does not have cirrus logic drivers, because of some strange bug i cant use harddisks on qemu on arm
<shivansps>
its translating the .so libs, those libs are compiled by the NDK
<shivansps>
it has problems translating native C...
<shivansps>
but they maded because skype ended up working on a version
<shivansps>
as well netflix
<shivansps>
but stopped working again after ap update haha...
<shivansps>
but MS is more than capable of doing a similar and decent system...
<shivansps>
they just dint want to... they want to sell you new apps
<shivansps>
and lot of these ARM apps where games, with ARM .so libs, the app was capable for translating the .so libs fast enoght for the android games to be playable.
<bsdfox_>
are there schematics for a2000 anywhere?
<mnemoc>
no
<bsdfox_>
also, why use a battery charger IC?
<RaYmAn>
axp209 is used for almost all a10 devices - it's part of the standard design
<mnemoc>
the axp209 is the official companion of all sunxi chops
<mnemoc>
chips
<RaYmAn>
iirc, i've only heard of mk802 dropping it
<RaYmAn>
with poor results =P
<bsdfox_>
ok. I'll see if I can probe it.. I bet it's just the charging LED and it looks like i2c is connected
<mnemoc>
olimex dropped it in their a13-micro.... no idea how it works
<mnemoc>
techn_: thanks
rm is now known as rinku
rinku is now known as rm
<lkcl>
there was another product that used 3 SY8008B LDOs. they also removed the HDMI RF shielding resistors, and wondered why people were blowing up the HDMI port
<lkcl>
it was a competitor to the Mele A1000
<lkcl>
someone mentioned it on the arm-netbooks list oo.... 6 months ago?
<mnemoc>
thats the mk802 iirc
<lkcl>
mnemoc: ahhh
<lkcl>
ok. oh dear :)
<RaYmAn>
it works vaguely ok
<RaYmAn>
but there was/is huge issues with current kernels
<RaYmAn>
like, dynamic cpu freq stuff makes it crash
<mnemoc>
i think turl fixed that
<RaYmAn>
okay, cool
<RaYmAn>
my mk802 has..well..gone into unused mode mostly.
<mnemoc>
:)
<RaYmAn>
I haven't managed to get proper output from it yet - messed up colors when using hdmi->dvi adapter
<mnemoc>
RaYmAn: if you have the chance, please test it with the current stage branch. it might fix stuff
<RaYmAn>
I'll try and look at it tomorrow
<mnemoc>
we recently merged some "fixes" from the a13 sdk related to csc and hdmi
<mnemoc>
seems to improve performance too
<RaYmAn>
cool
<mnemoc>
at least empat0 claimed around 10% better performance after the merge of the "csc" (whatever that is) merge
<Turl>
mnemoc: I'll have to give that a try then :)
<mnemoc>
current sunxi-3.0 includes it. obviusly testing the stage is preferred
<Turl>
the main issue with lacking AXP is that you don't get DVFS
<mnemoc>
Turl: people on the issue tracker said one of your commits fixed that...
<Turl>
eh
<Turl>
as far as I understand it, AXP does the voltage scaling
<Turl>
so DVFS without an AXP becomes DFS right? :P
<techn_>
mnemoc: there is now couple new commits
<bsdfox_>
32 关机、电池检测、CHGLED 控制寄存器 R/W 46H
<bsdfox_>
anyone know chinese and can decipher the i2c address?
<penguin42>
no, but I'd guess 46 Hex
<RaYmAn>
tried google translate?
<bsdfox_>
penguin42, that's just the register we want to change
<RaYmAn>
It's so tempting to reply to that post, but it clearly won't help.
<mnemoc>
better to ignore and keep pushing ahead
<Turl>
indeed
<RaYmAn>
yeh - the thing that annoys me is mostly the lack of informed facts
<RaYmAn>
replacing broken support for generic a10 with correct support for specific devices isn't really a step back
<Turl>
+1
<Turl>
besides adding in your device is rather trivial
<RaYmAn>
though, couldn't we theoretically add the old "generic" settings for generic vaguely working spl support? at least for a10
<mnemoc>
the sun4i branch has the old hardcoded stuff
<Turl>
RaYmAn's idea sounds sensible
<Turl>
make a new generic-sun4i board with the old hardcoded settings
<mnemoc>
that's mele_a1000
<mnemoc>
the hardcoded data was mele's data
<Turl>
needs to be documented then :)
<Turl>
wiki wiki
<RaYmAn>
^^
<mnemoc>
and the old sun5i's spl was a13_olinuxino
<techn_>
:D
<RaYmAn>
I got the point that the data was the same, but it didn't quite dawn on me that it means to just use mele_a1000 target => generic support
<RaYmAn>
(generic support in the sense of working as well as previous versions)
<mnemoc>
but 512M-onlu
<mnemoc>
but 512M-only
<RaYmAn>
sure, but still
<mnemoc>
yes
<hno>
techn_, there was zero meaningful discussion regarding u.-boot from a10linux team. And also close to zero stuff actually done other than a modularized default environment.
<hno>
I have pushed their u-boot branches to my repository now.
<techn_>
hno: no doubt.. but I think it would be useful to make correction to that forum.. so that random surfer wont get impression that this is the situation
<hno>
find it quite hillarious to read that discussion "Linux-sunxi developers (primarily Android devs it seems) fail to deliver adequate Linux support in a timely mannner. They seem only to be in it to help Tom Cubie and Olimex to sell devices." and "LInux-sunxi is there for the profit makers.". what the hell?
<mnemoc>
profit?
<hno>
The wise words of gnexus.
<Turl>
they also claim somewhere there that we move slow and we should've gotten mainline support already :P
* Turl
starts making pastelitos to make mnemoc hungry
<hno>
Turl, in same thread he also says "after careful discussion between the 3 founding community members, we regret to inform that the a10linux.org project has been suspended" followed by " In the words of the guy who has the hosting account, and made the "Guiness-inspired decision" to "rm -rf *" everything: " . "You can thank him for deleting all the Linux images, etc. . . I did not have the space on my laptop to make backups. "
<RaYmAn>
that's always the kind of people you want running things ;)
<Turl>
hahaha
<RaYmAn>
wtf
<RaYmAn>
"If somebody had offered free hosting, and at least one or two other developers or project leads from other sites had joined up, we would have kept it going. "
<RaYmAn>
so basically, if they had found people to pay for and do all the work, they would have kept it going? :P
<mnemoc>
shame on you lundman
<Turl>
RaYmAn: reminds me of some android communities
<RaYmAn>
yeh
<RaYmAn>
Turl: and in particular the android "developers" going all "I didn't get enough donations for adding shitty hacks to other peoples linux kernels so I decided to go FOR PAY and not share sourcecode"
<mnemoc>
i have logs to show I did offer gnexus to join us when he came to announce "the" a10 community, a10linux.org thing
<mnemoc>
actually, ibot was present too :)
<RaYmAn>
heh
* RaYmAn
shrugs
<RaYmAn>
I think they are a lost cause
<rm>
hno, he had nothing useful there anyway
<rm>
"minimal" images weighting 4 GB
<rm>
with everything and a kitchen sink installed
<rm>
from my impression he was very enthusiastic but just way too incompetent
<mnemoc>
no. but I found a nice deal in hetzner's used offers and I'm upgrading the server where I host linux-sunxi.org to be able to have an lxc for hosting and another for building
<RaYmAn>
lxc? xen ftw :P
<mnemoc>
i don't need separated kernels or virtualization at all
<RaYmAn>
:)
* hno
have only bad experiences of xen.
<mnemoc>
only separation and resource limits
<RaYmAn>
hno: really? like what?
<hno>
Linux dom0 drivers failing under Xen on every box I tried on.
<RaYmAn>
odd. I've never had any issues with my xen box
<hno>
plus completely destroying any power management.
<mnemoc>
ouch
<RaYmAn>
power management? I don't think xen is quite made for that, so yeah.
* Turl
confused lxc with lxr and was wtf'ing a bit :)
<mnemoc>
lxc is basically vserver/openvz made upstream-friendly
<mnemoc>
or a secure chroot :p
<RaYmAn>
I guess my main beef with that is more related to providers overcomitting than the actual tech
<Turl>
openvz is kinda crappy
<Turl>
can't load custom modules
<Turl>
many iptables rules fail
<Turl>
tun/tap.. forget it :<
<RaYmAn>
I like xen cause I can keep the vms entirely seperated
<Turl>
RaYmAn: try kvm
<hno>
lxc is not quite ready for vserver hosting yet.
<Turl>
even better :)
<RaYmAn>
Turl: meh, doubt it :P
<RaYmAn>
also, I have a Widnows VM as well.
<Turl>
RaYmAn: with libvirt is effin' easy to use :)
<Turl>
RaYmAn: yeah kvm can run windows
<hno>
KVM runs Windows VMs just fine.
<RaYmAn>
ah
<libv>
argh! libvirt!
eFfeM has quit [Quit: Leaving.]
<RaYmAn>
I'm not too fuzzed about managing it manually tbh
* libv
has two things to add to his regexp filter
<penguin42>
Turl: I'd mostly agree, although still not quite as easily as some of the commerical stuff
<mnemoc>
libv: :D
<hno>
libv, make it a word based match.
<Turl>
penguin42: with virt-manager it's all point-and-click-y even
<RaYmAn>
PvOnHVM is quite neat for windows =P
* hno
tried using vmware. A nightmare.
<libv>
hno: it's become quite complex over the last 10y :)
<penguin42>
Turl: Yeh I use virt-manager, but I find it quite clunky; not very flexible, and absolutely useless at reporting what goes wrong
* Turl
+1s hno's experience
<Turl>
penguin42: it worked fine for me, to create a base debian image in minutes
<Turl>
then I scripted COW-images and launched 20 debians :)
<penguin42>
Turl: Yeh, same here - but things people like on vmware and some of the oracle one (erm what's its name) - like copy&paste from the guest, or mounting host filesystems etc, or plugging in USB devices - they're not simple with something virt-manager; and getting spice to work for fast graphics is pretty hairy
<bsdfox_>
hno, xen is good in the right environment
<bsdfox_>
I manage several production dom0
<Turl>
penguin42: yeah, kvm isn't aimed at the 'consumer market'
<Turl>
you're better off using virtualbox for that
<Turl>
kvm is well suited for running servers
<Turl>
or separated services
<mnemoc>
tried livesuit on kvm. no luck :<
<penguin42>
Turl: that's what I was thinking of....I don't think why kvm can't do that stuff, it just needs some care on the UI and some more glue
<Turl>
mnemoc: livesuit on virtualbox works fine
<mnemoc>
virtualbox, it's owned by oracle... yuck. no way
<hno>
bsdfox, quite likely. Just not sure it's better than kvm.
<Turl>
mnemoc: it's free software
<mnemoc>
Turl: it's oracle
<Turl>
mnemoc: so is mysql :P
<mnemoc>
mariadb :D
<Turl>
mariawhat? :P
<mnemoc>
community driven fork of mysql, works as dropin
<mnemoc>
i still can't recover from the impact of trying to learn intel asm back then.. ~12yo.... 20+ years ago
<mnemoc>
i know i can learn arm asm, but my brain refuses it
* hno
is not compatible with Intel asm. But managed to make a m68k to 80286 translator.
<mnemoc>
nice :)
<Turl>
movl $trauma, %eax
<Turl>
ret
<hno>
ARM asm is far easier to understand, but odd.
<Turl>
I heard MIPS one is nice to work with
* rm
puts Turl into a delay slot
<Turl>
xchgl %eax, %eax
<hno>
delay slots were fun. But best used by a compiler.
<hno>
and compilers do not need delay slots.
<hno>
and did not work out to great when complexity of the cpu increased.
<rm>
there aren't any 'naive' direct execution MIPS CPUs anymore since a long long time, right
<rm>
the dream was, I guess, to have those
<rm>
+ have them running at 10 GHz
<rm>
or something like that
<hno>
rm, what is a " 'naive' direct execution" unit?
<rm>
hno, I meant ones that are not superscalar and not out of order and do not pre-decode MIPS instructions to some other internal instruction set before actual execution
<rm>
and not predicting branches, etc
<hno>
ah, no not that I know of.. They all do those things which nullify the benefits of the delay slot.
<rm>
as you noted delay slots were designed with the simplest CPUs like these in mind
<Turl>
it looks like it's a trick to work around a design limitation
<hno>
Turl, the idea is neat, and initially worked out well. But do not scale as frequency increases.
<hno>
the alternative is out.-of-order execution, which adds a lot of complexity to the CPU, but can scale further (with more complexity at the cpu)
arnd_ has quit [Ping timeout: 255 seconds]
<Gumboot>
ARM do not have NOP?
<RaYmAn>
not explicitly at least =P iirc, there is a particular syntax of LDR (i think) that's usually used as no op
<Gumboot>
Delay slots get the branch exposed earlier in the instruction stream so that prediction can start work sooner. There's not _really_ any pipeline-specific lock-in, but it does become a bit of an emulation problem and can fail to present any value whatsoever on more advanced pipelines.
<Gumboot>
I think there's a whole class of NOP operations, and one of them is guaranteed to be a pure NOP while others are just hints.
<rm>
branch prediction causes fetches from an invalid address
<hno>
they do branch prediction before brach target location is known?
<Gumboot>
I own one of those.
MindBeat has joined #arm-netbook
<techn_>
ZaEarl: Thanks for gt811 driver! I got touch working on a13_mid :)
<ZaEarl>
most excellent
<ZaEarl>
has it been pushed to github?
<techn_>
Not yet.. I'll send patches to ML.. to linux and u-boot
<techn_>
but that goes to tomorrow
<techn_>
how I can enable virtual keyboard on linaro?
<techn_>
linaro alip I mean
tmerle has quit [Ping timeout: 245 seconds]
<ZaEarl>
techn_, if you could grab the dmarg06 driver and make submit a patch for adding that to the repo, that would be great.
<techn_>
ZaEarl: sure
<ZaEarl>
I'll try to get some more drivers next week.
<techn_>
But I cant verify that anytime soon
<techn_>
.. or is there good way to verify g-sensor in linux ?:p
<ZaEarl>
I can test it.
<techn_>
.. you propably ment dmard06
<ZaEarl>
I just don't have my build compiling yet
<ZaEarl>
yes
<techn_>
yep.. my stock android has same sensor
<ZaEarl>
oh, good
<techn_>
do you have same tablet then?
<ZaEarl>
doubt it
<mnemoc>
techn_: can you turn that gt811 thing into a patch?
<ZaEarl>
my tablet has ft5x ts
<techn_>
I was thinking to get a13_mid into our sunxi-bsp system
<mnemoc>
techn_: sure, sunxi-boards
<techn_>
mnemoc: .fex file is in sunxi-boards.. u-boot, ts and gsensor atleast is missing
<mnemoc>
damn hetzner hasn't activated my new server yet :<
<mnemoc>
techn_: ts and gsensor goes to kernel.
<mnemoc>
for u-boot we can make an script
<techn_>
but I have u-boot and ts pending tomorrow on my machine ;)
<mnemoc>
:)
<techn_>
we should try to get i2c probing working for these drivers
cat_x301 has quit [Ping timeout: 252 seconds]
<techn_>
I had to manually add that ts driver to modprobe.d :(
<hno>
techn_, what's coming for u-boot?
<techn_>
hno: I was thinking to send board configs for mini-x and a13_mid
<hno>
nice
<hno>
did you dump boot1 header on those?
<techn_>
hno: what is preferred copyright for those?
<mnemoc>
hand over to denx?
<techn_>
hno: no.. params are from .fex :(
<hno>
ok. hope they are correct.
<mnemoc>
.fex is decompiled
<mnemoc>
it's data without copyright
<hno>
there is console pads on mini-x.
<hno>
also need data from some mk802 devices.
* mnemoc
still dreams with a tool to push over fel to copy boot0/boot1 into fel-readable memory
<hno>
mnemoc, yes..
<techn_>
hno: could you do dummy u-boot which dumps result to FAT drive?
<merbzt>
hno, andoma should be able to get you the data from mk802
<mnemoc>
hno: can one do sub-tools (like spl) from u-boot tree? to avoid duplicating....
<techn_>
In theory that should work? :/
<RaYmAn>
mnemoc: uboot standalone apps
<RaYmAn>
is probably what you want?
<mnemoc>
yes, sounds like it
<RaYmAn>
look in examples
<hno>
mnemoc, spl is not a sub-tool. It's a special build of u-boot.
<hno>
make spl/u-boot-spl
<hno>
but tree needs to be configured first.
<hno>
./mkconfig -A cubieboard
<Turl>
mhm, zfs
<hno>
what about zfs?
<mnemoc>
but once sunxi learns to do nand, one could make an "uboot standalone app" to push over fel and grab boot0/boot1, right?
<Turl>
I got it built on my desktop :P
<mnemoc>
hno: or it's better to make a new baremetal app?
<hno>
mnemoc, once we learn NAND it's easy to make a fel app for grabbing bootblocks.
<mnemoc>
:)
<hno>
It does not look complex. But too little time.
<mnemoc>
what happened to the guy who was interrogating you about the nand? maybe he can do it...
<techn_>
mnemoc: I had same in my mind :D
<techn_>
lol
* mnemoc
tries to remember his nick
<techn_>
gnexus? :)
<mnemoc>
:|
<techn_>
.. Or should we add script which modifies u-boot according .fex files
<techn_>
which sound ugly
<hno>
okt 24 22:05:03 <slapin_nb> hno, could you explain the role of sram with a10's nand controller?
<mnemoc>
next time I see slapin I'll ask him
<hno>
It was a good discussion. Made me understand the NAND controller much better.
<mnemoc>
me needs it for his goal anyway
<mnemoc>
s/me/he/
<ibot>
mnemoc meant: he needs it for his goal anyway
shivansps has quit [Quit: Page closed]
<hno>
what is his goal?
<mnemoc>
use a bare nand filesystem
<hno>
right
<mnemoc>
iirc he mentioned jffs2
<mnemoc>
but not sure
<mnemoc>
or it was ubi? .oO
<hno>
jffs2, ubi, etc. all require MTD
<mnemoc>
anyhow, it's on a post-it now and next time I see him in the channel I'll try to convice him
<hno>
and he did mention making an MTD driver
<hno>
okt 24 14:05:13 <slapin> hno: I need to make u-boot and kernel MTD to work with A10's NAND controller, a problem I face now is how to directly control NAND chip with this controller, so put NAND command, put data, get response, on low level, so actually how to put data byte on NAND. This is needed for some operations required by MTD, and for me to understand how it really works.
<mnemoc>
bootblocks grabbing via fel sounds like a perfect test case to learn to use the nand controllers :)
<mnemoc>
controller*
<mnemoc>
going to sleep now. good night
<hno>
Argh.. why did you post that forum link. gnexus is not a kind person.
<mnemoc>
don't read the troll
<mnemoc>
it's unhealthy
<hno>
I know. PLain accused me of pulling crap and making u-boot deviate even further from upstream.
<hno>
"I won't continue to waste time to support people (hipboi & hno) that pull this kind of crap and make things difficult for people who want control of their hardware." "The more work that gets done on hno's the more it deviates from standard Denx"
<mnemoc>
close that tab
<mnemoc>
it's just a pile of non-sense
<lundman>
heh
<hno>
NAND driver was pulled by Tom from u-boot because it's very ugly done and needed to be redone to match u-boot expectations.
<Turl>
don't feed the troll any longer :)
<techn_>
nam nam nam.. troll is very hungry :(
* Turl
feeds techn_ with disp driver
<techn_>
Turl: ? :)
<Turl>
(20:40:11) techn_: nam nam nam.. troll is very hungry :(