<libv>
should be able to work off of 3V too according to its datasheet
<specing>
do you get 3v3 on the tx rail on the mele?
<libv>
i did last week, taking the multimeter to it again
<specing>
then use my spamming program and send 0 to it, the rail should read 0 - 0.3V
<specing>
*from it
<libv>
voltages are all correct, all are ~3.2V above gnd at this point
ZaEarl has quit [Quit: Ex-Chat]
<libv>
i just spent another 15EUR on ebay, will give things another chance when those bits arrive
ZaEarl has joined #arm-netbook
rellla has joined #arm-netbook
<specing>
-.-
<specing>
libv: did you try sending 0 from mele?
<specing>
and measuring?
kairu0 has joined #arm-netbook
<libv>
specing: i am seeing some form of characters fly past when things are correctly wired up, and they definitely are plentiful when booting
<libv>
and when i type stuff, things happen too
<libv>
it's just garbage characters
madmalkav has quit [Quit: Leaving]
<libv>
but no, i haven't tried it, and tbh, i would rather do something more productive like reducing a 56MB mali memory dump to find out what i am missing in my parsed command stream for a q3a frame
<libv>
next week i will give it another go
<libv>
when the new bits have arrived
<specing>
ok
rellla has quit [Quit: ChatZilla 0.9.88.2 [Firefox 13.0.1/20120614114901]]
naobsd has quit [Ping timeout: 245 seconds]
Quarx has joined #arm-netbook
ZaEarl has quit [Quit: Ex-Chat]
<Triffid_Hunter>
libv: those pl2303s are crap, get an ftdi
<libv>
Triffid_Hunter: it worked reliably for a year and a half, and still happily works on my n9
<specing>
Triffid_Hunter: pl2303 has worked for me for the past 5 years
<specing>
they sure cant go over ~115200 but they work
<specing>
Im going to build myself a FT232H (serial+JTAG+...) board when I start doing os-less ARM dev
<RaYmAn>
the pl2303/cp2303 chips are awesome for what they provide - cheap as hell, works for regular 115.2k serial. It's nice not having to care that much if you accidentally fry the ttl-USB chip ;)
eebrah has quit [Quit: ChatZilla 0.9.88.2 [Firefox 10.0.5/20120606062003]]
lkcl has joined #arm-netbook
rellla has joined #arm-netbook
<mnemoc>
hno: please page me when you have a minute to talk about memory info. It's too much new knowledge for me :( .... in my mele the fixup is currently receiving a meminfo with zero banks, and an atag of size 0x20000000 at 0x40000000. their hack shrinks at to 0x1c000000
<mnemoc>
somehow I was expecting the meminfo to come prepopulated
<mnemoc>
and have no clue how the kernel is supposed to know to isolate those 64M from the first bank when going hack-less
<mnemoc>
hno: should I remove the 64M from the first ATAG_MEM or just assume magic(tm) will solve it?
naobsd has joined #arm-netbook
eebrah has joined #arm-netbook
<RaYmAn>
so, anyone considered the ouya as a replacement mele type thing? Though, not sure exactly how hackable it is
<mnemoc>
it's tegra... isn't it? :(
<RaYmAn>
yeah
<RaYmAn>
linux support is still far better than a10 ;)
<RaYmAn>
not open source graphics stuff though
<mnemoc>
:)
<mnemoc>
considering the nightmare it was for you to get a partially working adam/ics *with* notion ink support I wouldn't consider any tegra device hack friendly
<mnemoc>
of course they have the advantage of a larger head-desking^Whacker community and sexier devices
<RaYmAn>
tbh, the biggest issue with adam was just how fucked up nvidias 2.6.32 framework was
<RaYmAn>
from what I've heard from other devices developers, most simply dumped 2.6.32 and reimplemented eveyrthing from scratch for 2.6.36
<RaYmAn>
given we couldn't do that, it was a huuuge undertaking :P
<RaYmAn>
now that we have 2.6.39 and 3.1 kernels, it's *much* easier. They actually use standard interfaces now :P
<mnemoc>
:D
* mnemoc
wants JB/3.1 on his adam!
<RaYmAn>
heh
<RaYmAn>
it's easy
<RaYmAn>
we're almost done with JB on TF201
<mnemoc>
adam's cm9/b5 is unsuably slow (unless you are playing 3d games, which ain't my business)
<RaYmAn>
weird
<RaYmAn>
Tbh, I've mostly been involved in kernel side
<RaYmAn>
it's not one of the versions where the PMU spams IRQ's sufficiently to slow down system, is it? :P
<mnemoc>
haven't researched... I only use it to read ebooks... but any task involving nand access puts it on it's knies
<mnemoc>
all processes halt
<RaYmAn>
weird, afaik there should be no changes to nand in any recent versions
<RaYmAn>
but anyways, porting to 3.1 is fairly trivial
<mnemoc>
go go go :)
<RaYmAn>
We tried to keep the mods to non-board files as low as possible, so you can mostly just copy board-adam* across, edit makefiles and get boot
<RaYmAn>
then switch out android gralloc/hwcomposer with updates ones
<mnemoc>
great! then let others take care of the transformer and bless the zombie adam community with a JB/3.1 :)
<RaYmAn>
lol
DonkeyHotei has quit [Excess Flood]
<RaYmAn>
It's more fun to work on TFP :P
<mnemoc>
:(
<RaYmAn>
With it's multiple viewing angles and stuff
<RaYmAn>
:P
<mnemoc>
:)
DonkeyHotei has joined #arm-netbook
<mnemoc>
it's sad how NI wasted their momentum
<mnemoc>
and it's community
<RaYmAn>
yeah
<RaYmAn>
They are still alive though
<RaYmAn>
lol
<mnemoc>
doing what?
<mnemoc>
or just... alive?
<RaYmAn>
working on adam2 :P
<mnemoc>
their adam2 blog is even more full of BS than their original blog :< .... and also stalled
<RaYmAn>
yeah, they stopped posting for a bit
<RaYmAn>
but they are still working on it.
DonkeyHotei has quit [Ping timeout: 248 seconds]
<RaYmAn>
they are keeping things "secret" now, so they can actually show up with a working product, lol
<mnemoc>
:D
<RaYmAn>
I think they've learnt from their mistakes ;)
<RaYmAn>
More stuff open source from day1 and such, lol
<mnemoc>
yeah, very
<lundman>
:)
<lundman>
cd
<RaYmAn>
mnemoc: tbh, that ouya is almost same price as mele and quadcore :P hehe. But yeah, I dunno how open it actually is
<mnemoc>
RaYmAn: afaik the ouya still isn't anything. the mele is here ;-)
<rm>
I read "Tegra" and close the browser tab
<rm>
so I don't even know, e.g. does it even have SATA?
<rm>
Ethernet?
<rm>
how much RAM, etc
<RaYmAn>
sata is unlikely
<RaYmAn>
ram etc is specified on site.
<rm>
iow, Mele is still massively more attractive
DonkeyHotei has joined #arm-netbook
<rm>
the only thing I see going for Ouya, is supposedly a more powerful GPU
<RaYmAn>
and a much more powerful cpu? :)
<RaYmAn>
but yeah
<RaYmAn>
iirc, it's 1gb ram, 8gb flash, they are looking at adding ethernet
<rm>
But does it run GNU/Linux? (c) :)
<mnemoc>
but as there is no ouya yet, we have time to see what other A9 alternatives come out
<RaYmAn>
rm: that's the question. There are relatively decent X11/OpenGL ES drivers available, even in hardfp versions, but it depends on how open it is
<RaYmAn>
mnemoc: true
<RaYmAn>
Cortex A15 stuff next year :]
<mnemoc>
yeah
<mnemoc>
no rush in trashing more money :p
<RaYmAn>
indeed
<RaYmAn>
I'm kind of battling FOMO ;)
<RaYmAn>
(fear of missing out)
<mnemoc>
:)
<rm>
hmmmm
<rm>
so it doesn't even have ethernet yet
<RaYmAn>
but otoh, I already have one tegra3 device with another on the way, so I really shouldn't buy anymore stuff liek that, lol
<RaYmAn>
It's depressing to realize I have three tegra2 devices, two tegra3 devices, and various others :/
<lundman>
might consider that odroid-x.. might work out better than a10
<rm>
define work out and define better :p
<lundman>
well, work out better for me I suppose. :)
<lundman>
mele does not do any audio, so I cant use it for mediaplayer
<mnemoc>
it does for me...
naobsd has quit [Ping timeout: 245 seconds]
<rm>
hmmm, never tried audio on Mele yet
<lundman>
can do stereo to tv, but plug it into amp and it stutters bad
<lundman>
and cant go more than stereo
<rm>
RCA?
<lundman>
the spif doesnt turn on
<lundman>
not tried rca, want > 2 ch
<mnemoc>
I have one mele a1000 taken apart on my desk, but another with Tom's ICS on the living room connected to the tv playing music videos almost all day...
<mnemoc>
ok, I use hdmi
<lundman>
great, play a movie with good sound
<lundman>
5.1 or something
<rm>
btw, is that all on Android?
hackandfab has joined #arm-netbook
<lundman>
all android
<lundman>
and I dont mean play a movie with 5.1, and listen on a stereo tv. connected to any avr
<rm>
I am surprised it would have problems on Android
<rm>
but for my uses I'll be happy enough if I get 2ch over RCA (in Debian)
<mnemoc>
they seem to have removed fancy sound decoders from the official image because of licensing problems
<mnemoc>
but that doesn't mean "it doesn't do sound"
<hackandfab>
hello all, I'm new here, do you have a FAQ?
<lundman>
decoding means going to 2ch
<lundman>
it should do passthrough, both HDMI and spdif
<L84Supper>
when TI released the 4430 and the pandaboard they said in 6 months to 1 year that parts would be available through distis and not direct
<traeak>
yuk powervr
<L84Supper>
it's 1+ years and still no parts available through distis
QingPei has quit [Ping timeout: 244 seconds]
<Turl>
rm x86 AIOs have touchscreens
<Turl>
on my experience they resemble resistive screens more, and are rather crap to use
<Turl>
I'm unsure on what's the actual touch technology in use
<traeak>
doesn't it bother the crap out of you when people touch your screen?
<traeak>
fingerprints, etc on workstation screen is really really bad
<Turl>
I don't notice that kind of stuff
<Turl>
the only time I notice it is when I clean my screen while the PC is on
<rm>
I suppose with touch screens it will be natural that they're all in fingerprints
<hno>
traeak, you need automatic winshield wipers on the screen.
<rm>
so you won't notice
<Turl>
then I'm like 'holy crap it was dirty, the screen looks like a new 4K monitor now'
<rm>
with pristine non-touchscreens, of course a single fingerprint of an annoying person who just came by, stands out...
CoconutCrab has joined #arm-netbook
<Turl>
rm I use a non-touchscreen 16" laptop
CoconutC1ab has quit [Ping timeout: 240 seconds]
QingPei has joined #arm-netbook
ka6sox is now known as ka6sox-away
<L84Supper>
anyone hear if the secure boot got cracked yet for the TI omap44xx? or maybe they just worked around it
DonkeyHotei has joined #arm-netbook
kaspter has quit [Ping timeout: 246 seconds]
<Shivan>
what is the drm.h file that is requiered to compile mali driver? i dont remember part of what it is... wrhombus wiki says only ump includes are needed but the driver also ask for several drm*.h files in usr/include
tom_say has joined #arm-netbook
gimli has joined #arm-netbook
CoconutCrab has quit [Quit: lết vào chăn]
<mnemoc>
Shivan: afaik it's from mali's DDK, which is NDA'ed
<Shivan>
But i was able to compile them with another distro before wiping the sd... strange...
<hno>
drm/drm.h is a standard kernel include file.
<hno>
should be installed by default if you have C/C++ development support installed (usually kernel-headers or similar package).
<hno>
And also found in the kernel includes.
* mnemoc
scratches his head.... .oO(which was the header from the DDK hipboi had to share?)o
<hno>
UMP?
<mnemoc>
possible... but I had the idea it was something *dr[mi]*
<mnemoc>
hno: did you see my question about meminfo this morning?
<hno>
no
<hno>
can you repeat it?
<mnemoc>
basically how do we deal with the 64M removed by fixup from bank 0
<mnemoc>
should I mangle the first ATAG_MEM, or trust "magic" will happen without intervention
<hno>
Peek at other MALI implementations how they reserve the shared memory?
<mnemoc>
ok
<mnemoc>
wasn't sure the owner of those 64M
<mnemoc>
and another question, do you know if their bootloader passes ATAG_MEM? if they don't it would be a single way to have a fallback
<mnemoc>
simple*
<mnemoc>
will probably need to boot a debugging kernel from nand and see..
<hno>
have not looked at what the allwinner boot loader passes to a kernel. The mele chainloads u-boot.
<mnemoc>
in which case I will receive dummy ATAG_MEM, right?
<mnemoc>
and so, no fallback possible :<
<mnemoc>
meh
<hno>
chainloded u-boot gives a dummy memory information atag yes.
<hno>
but easily fixed to build proper memory info.
<mnemoc>
CONFIG_SUNXI_IGNORE_ATAG_MEM default n ?
<hno>
Proposed config for disabling the override?
<mnemoc>
yes
<mnemoc>
if someone wants to use allwinner's broken bootloader can enable that...
<hno>
Or just upload a fixed u-boot next to the kernel.
<mnemoc>
meaning you'll fix the nand branch? :)
<hno>
me or someone with a 1G system.
<hno>
regarding MALI DDK, the A10 SDK MALI binary do include a bunch of headers: http://fpaste.org/nIE2/
<hno>
Mix of SGI Open Source licensed files and the more important ones "This confidential and proprietary software may be used only as authorised by a licensing agreement from ARM Limited"
<hno>
No idea what that licensing agreement says
<mnemoc>
we received the binaries from allwinner...
<mnemoc>
hopefully one day linaro will convise ARM of stopping the BS
<hno>
also iirc the address is hardcoded in the MALI driver.
<mnemoc>
I was wondering about A10 systems with less than 512
<hno>
if I had an HDMI display...
<hno>
have not seen any with 256M.
<hno>
probably because ~128 IS AVAILABLE THEN.
* hno
releases caps-lock
<xxiao>
anybody is able to replace u-boot on minipc
<specing>
mnemoc: haha, unit price rises the more units you buy ;P
<mnemoc>
specing: haha... missed that detail
<mnemoc>
hno: a good thing of having mali in-tree is that we will be able to release those 64M when mali is not enabled ;-)
<hno>
mnemoc, and related it means anyone who disables mali for building out of tree gets seriously burnt
<mnemoc>
hno: meh
ka6sox-away is now known as ka6sox
lupul has quit [Ping timeout: 245 seconds]
Quarx has quit []
<rm>
mnemoc, $60.99-$5 on Aliexpress today
* mnemoc
wants a mele with 1GB :(
<hno>
tempting.. if we knew UART0 was accessible without too much effort without the need for sd breakout.
* hno
is still dependent on SD.
<mnemoc>
it has an uart directly next to the soc
<mnemoc>
soldering required, but that isn't a problem for you
<mnemoc>
hipboi_: please push mele for a 1GB model.... that could really be used as desktop
<DonkeyHotei>
does he have sway with mele?
<mnemoc>
he is a direct buyer...
<mnemoc>
meaning he has access to sales people inside mele.cn
<DonkeyHotei>
but how much power do they have over that?
<hno>
mnemoc, yes it should be there, but I have not seen anyone confirming RX works.
<mnemoc>
i bet it's just an incomplete script.bin
<hno>
DonkeyHotei, none.
<DonkeyHotei>
thought so
<hno>
And I do doubt the next 1GB Mele "livingroom PC" will be A10 based.
<mnemoc>
more than us
<DonkeyHotei>
unless it's A15
<Shivan>
probably the am dual core
<mnemoc>
as in sun6i or cortex-a15?
<specing>
hno: heh, scope
<DonkeyHotei>
sun6i
<traeak>
market's moving too fast, eh?
<Shivan>
i dont think there is a chance for a mali t605 on next allwinner no?
<DonkeyHotei>
traeak: always
<specing>
Shivan: there is
<specing>
Shivan: Im sure there will be on a quad-core model
<Shivan>
im hope so... full OpenGL and OpenCL support its a dream
<Shivan>
IF WORKS
<Shivan>
i still cant make es2 to work for my mali400 -.-
DonkeyHotei has quit [Ping timeout: 255 seconds]
<rzk>
cool, on low-power a10 throws kernel panic then reboots to boot1 for a picture of low-battery. i thought it will just hang hard till next manual boot.
<rzk>
also, hv2605 appears to be broken, its a capacitive buttons driver, throws infinite amount of "incomplete xfer (0x28)" to log.
rellla has quit [Quit: ChatZilla 0.9.88.2 [Firefox 13.0.1/20120614114901]]
<mnemoc>
patches VERY welcomed :)
<mnemoc>
but a ticket is a good first step
<hno>
rzk, you have matching hardware for hv2605?
<hno>
mnemoc, actually, I retract that. Platform driver registration and memory reservation need to go hand in hand, separate from the MALI driver itself.
<mnemoc>
hno: please rephrase
nibb_ has quit [Read error: Connection reset by peer]
nibb_ has joined #arm-netbook
<mnemoc>
hno: I fail to understand that sentence :<
<hno>
mnemoc, see arch/arm/mach-sun4i/devices.c
<mnemoc>
the drm device?
<rzk>
seems so, hv2605 probe detects it. right after probe done xfer stuff happens. I'll check with android init.rc. I've loaded other drivers for keyboards/keypads (sun4i-ts, sun4i-keyboard, sun4i-keypad) as well and tested for input at all /dev/input/eventX, only ft5x_ts sends data, but I think its just junk data because capacitive panels are close to each other.
<mnemoc>
hno: that was added by us to allow the x11 driver to work
<hno>
Hm... need to double check then.
<mnemoc>
my guts tell me it's a bad idea to try to keep mali decoupled from the kernel tree if we want proper support... but...
<mnemoc>
android doesn't need the drm device
<mnemoc>
only x11
<mnemoc>
but my first priority currently is to finish cleaning up plat-sunxi's sysconfig in 3.4
<mnemoc>
sys_config*
<hno>
There really should be a platform_device detailing the memory reservation to the MALI driver. But there isn't.
<hno>
It's not our job to fix the MALI driver. Only minimally working around it.
<mnemoc>
sure, only the hooking
<mnemoc>
no intention in messing inside the driver beside the "platform.c" file
<rzk>
did you know that FES has a nice ASCII logo inside? :3
<mnemoc>
uh
<rzk>
says "Efex" and some chinese name
<mnemoc>
the sata disk I connected to the mele sounds like a cricket...
<traeak>
what kind of speeds you getting ?
<hno>
Right, that mali_platform is mostly a dummy, only saved by pm/.. "by accident" doing the magics needed to shut down MALI on suspend by shutting down all clocks leaving MALI hanging in mid air.
<traeak>
mnemoc: run hdparm? :-p
<hno>
rzk, which FES? The main (bigger) one?
<mnemoc>
traeak: at 1GHz, ~60MB/s ... and sightly over ~300MB/s cached
<mnemoc>
at 60MHz around 18MB/s
<traeak>
mnemoc: what's the chance i somehow "broke" my mele ?
<traeak>
60mhz? is there some setting i'm missing?
<traeak>
ahhhh
<mnemoc>
traeak: ondemand governator ;-)
<mnemoc>
traeak: test using "performance"
<traeak>
yeah crap..lets try to figure out how to chnage that again :-p