<steev>
Turl: oh, well, i did the init, and then the local_manifest, but when attempting to sync, it complains that refs/heads/master isn't available
<Turl>
brunch again
<furan>
brunch again just gives me a huge list of options that does not include my device
<steev>
furan: . build/envsetup.sh
<furan>
I'm doing . build/envsetup.sh && brunch
<Turl>
furan: it's brunch <device>
<Turl>
to build that device
<furan>
derp
<furan>
thanks
<Turl>
or breakfast <device> if you just want to set the env up and use mm/mmm/make
<Turl>
/mka/eat/etc :P
<steev>
bacon
<Turl>
steev: that might be github failing
<steev>
make bacon!
<Turl>
mka bacon for auto-j + schedtool and ionice
<steev>
ah right, mka
<steev>
i always forget that
<Turl>
make bacon will work too
<steev>
Turl: how goes the jb work? or you stuck while waiting on new cedarx?
<steev>
i'd say try a linaro base, except the cortex strings that they replaced are optimized for a9, and not really tested on a8
nibb_ has joined #arm-netbook
<Turl>
steev: well, audio works, wifi works ok if you disable the new 'weak wifi' thing (idk why but my driver likes to report a crap signal while authenticating and android doesn't like it), and graphics work
<Turl>
cedar is not yet there, nor is camera which depends on cedar
<Turl>
and that's pretty much it for an a10 tab, not that much hw :P
<Turl>
sensors work too
<Turl>
sensor*
<steev>
nice
<Turl>
so does brightness
<steev>
switch ics to jellybean?
<steev>
or not public yet so as people can't shoot themselves in the foot
<Turl>
yeah I've pushed most if not all of what I have now
<Turl>
but idk, not having cedar and camera is kinda dealbreaker for some
<steev>
i can count on one hand the number of time i use the camera on my tablet
<Turl>
my tablet is mostly a reading device, and I check out one or two videos while reading tech news and such, so it's kinda a bummer not to have cedar
<furan>
hm. I did something silly. I ran the script to grab propriety files and it grabbed libmali.so but the build isn't finding it
<Turl>
yeah I don't miss the cameras at all
<steev>
(which is usually, <did it work? yep, cool> or <did it work? nope... oh well>
<steev>
yeah, that's about the only thing, lack of TED would make be a bit sad, but my Nexus 7 arrives tomorrow so... i get to go to town on the LY-F1
<Turl>
furan: check the lib is where it's wanting it
<Turl>
I might have foobar'd a path or two making the example repo
<furan>
yeah lookin
<steev>
bedtime
<steev>
night all
<Turl>
night steev
<furan>
'night
<furan>
files are in vendor/allwinner/ydpg18/proprietary/lib
kaspter has quit [Ping timeout: 240 seconds]
<furan>
I copied the proprietary files into my out obj dir. I know I'm not supposed to do that, heh
<furan>
want to see if it finishes building
<furan>
guessing it's just the name for the vendor library path is wrong
<furan>
how do you deal with a closed-source .ko without reverse engineering it?
<lundman>
I RE it
<furan>
ok
<furan>
I am pretty good with ida and arm assembler, have a small driver to move
<lundman>
yeah IDA would be my choice
nibb_ has quit [Read error: Connection reset by peer]
<furan>
it's silly. for the yinlips gamepad they took the 4xi ir driver and just modified it for the gamepad io, which appears to be a parallel to serial shift register that is attached to the ir i/o signals
<furan>
but then they read some different gpio as well
<furan>
not a huge driver, lot of shared code with the ir driver
<furan>
so shouldn't be too bad
kaspter has joined #arm-netbook
<lundman>
you are doing RE for a gamepad?
<furan>
yinlips ydpg18
<furan>
it's an a10 psp clone
<lundman>
ah
<furan>
the gamepad driver sucks (it has an analog stick, read as analog i/o, but then they gave up in the driver and just send dpad input with it)
<lundman>
heh
<furan>
plus to do my own rom (since I want to use larger nand) I have to reimplement it
<furan>
brunch compiles llvm and clang twice :(
<furan>
hmm, now it's failing to find RomManager.apk
<furan>
nevermind, running get-prebuilds
<furan>
t
nibb_ has joined #arm-netbook
Schnabeltier has joined #arm-netbook
<Turl>
furan: brunch is a full build
<Turl>
and no, you're not supposed to copy stuff manually to out
<Turl>
check the imports on device/allwinner/whatever/* that calls the vendor-something.mk and make sure it's ok
rzk has joined #arm-netbook
rz2k has quit [Ping timeout: 245 seconds]
<furan>
hrm
<furan>
I got a recovery.img and dd'd it to /dev/block/nandg
<furan>
but when I try to enter recovery boot (home+power+down on the yinlips) it doesn't enter it
rz2k has joined #arm-netbook
<Turl>
boots android? try again
<Turl>
otherwise plug to your pc and adb shell
rzk has quit [Ping timeout: 245 seconds]
<furan>
I also got no love from adb reboot recovery
<Turl>
maybe it's booted but just no screen
<Turl>
I suppose you have the LYCHEE LCD/DISP/HDMI stuff built in on your kernel? you should if you don't
<furan>
it keeps booting into android
<furan>
Id be happy if I got black :)
<Turl>
yeah it's possible adb reboot recovery doesn't work on stock rom
<furan>
so it's running ics now. thinking about flashing it back to 2.3 as I might get recovery from adb
rz2k has joined #arm-netbook
<Turl>
I'm asking because gb and ics partitions seem to be different
<furan>
you sure i dd'd to the right nand block?
<furan>
used nandg
<lundman>
nandc for regular kernel
<Turl>
ics has recovery on nandg, gb apparently has it on nande
<Turl>
nandc is regular kernel on ics
<furan>
so, being that it's impossible to brick an a10, since livesuite + a10 brom work together to do the flash
<furan>
it'd be fine if I wrote the recovery to nandc to see if it even works right?
<furan>
(good thing the kernel's not paged)
<furan>
ok well the good-ish news is when I put recovery on nandc, it boots and displays just dandy
<lundman>
I wrote my test kernel to nandc (using the original initrd) and when it failed, booted miniand to dd the original nandc
<furan>
ok I went about it in an extra foolish way. I booted to nandc recovery, and then flashed from there
<furan>
now when I boot I get a black screen
<furan>
which I am not surprised about
<furan>
adb shell works
<furan>
no root
<furan>
strange that it works fine w/recovery
<furan>
nothing interesting in dmesg
<Turl>
did you include disp/hdmi/lcd lychee stuff as =y on your defconfig?
libv has quit [Ping timeout: 272 seconds]
<furan>
yeah. I assume if it works in recovery it'll work in real boot
<furan>
maybe I assume wrong?
<Turl>
check you have the mali stuff in place too
<Turl>
run logcat and see if there's all the stuff dying
<Turl>
(adb logcat)
<furan>
lots of death
<furan>
I did get a warning that I was using the wrong java to compile
<furan>
(java 6 but not java 6 se)
<Turl>
check that you have the mali libs on system/lib/ and system/lib/egl/
<Turl>
yeah that's a non issue
<Turl>
OpenJDK works just as good
<furan>
wow no mali
<furan>
why did recovery work? it just use straight fb?
<furan>
let me push and see what happens
<Turl>
yeah recovery just writes to fb
<furan>
hrm. I guess I need to fix that whole path issue then
<furan>
for some reason I don't have root (my fault?)
<Turl>
furan: 'adb root'
<furan>
so I can't remount and write the files
<Turl>
then 'adb remount'
<furan>
disabled by system setting
<Turl>
hm yeah not much use if you can't get to the system :D
<furan>
hehe
<Turl>
furan: boot recovery and push the files there
libv has joined #arm-netbook
<furan>
if I can get into recovery :P
<furan>
I will likely reflash using livesuite and rebuild
<Turl>
with the issues linking you mentioned earlier, and now the fact that the files didn't get copied at all
<rz2k>
whats the best way to handle chinese symbols in C? I mean I have binary with mess of standard readable ASCII and some chinese codepage, need to store that info somehow
<Turl>
sounds like it's not importing the vendor makefile
<mnemoc>
Turl: the point of wip/3.0/mem is not to reduce the size allocated by mali but to reserve it correctly
<mnemoc>
Turl: modifying the meminfo is simply wrong
<mnemoc>
Turl: also (512 - CONFIG_SUNXI_MALI_SIZE) alters the start address for every value
<mnemoc>
Turl: which again will break the libraries for anything different than 64
Quarx has joined #arm-netbook
zerosheds has joined #arm-netbook
fullofSheds has quit [Read error: Connection reset by peer]
steev has quit [Remote host closed the connection]
Hexxeh has quit [Remote host closed the connection]
steev has joined #arm-netbook
gnexus has joined #arm-netbook
<gnexus>
mnemoc: tried playing with thumb2 kernel again just for fun.
<mnemoc>
killed your pets?
<gnexus>
mnemoc: dead. dead. dead!
<gnexus>
module directory loses the + in thumb2
<gnexus>
kernel loads but it doesn't like the suni stuff. . .
<gnexus>
Unhandled fault: alignment exception (0x801) at 0xf1c13041
<gnexus>
hcd controller kills it
<gnexus>
i would hate to attempt to fix all that!
<gnexus>
not worth it for 10% smaller kernel size. . .
<mnemoc>
don't you prefer to invest that energy in finding how to fix mali/memblock_reserve() support ? :)
<mnemoc>
wip/3.0/mem branch
<gnexus>
mnemoc: yes. i was just compiling anyway so i let it run overnight on the mele w/ thumb2 config
kaspter has quit [Ping timeout: 248 seconds]
<gnexus>
do still need to get the kernel smaller though
<mnemoc>
why "need"?
<gnexus>
2.2MB is still way too big
<gnexus>
mnemoc: smaller kernel size = faster system
<gnexus>
wouldnt matter so much if there was 2GB+ memory
<mnemoc>
have you profiled that claim?
<gnexus>
mnemoc: linaro and cnx have some articles on it
rellla has joined #arm-netbook
<gnexus>
most kernels are ~1.5MB
<mnemoc>
about relevant performance improvements for moving key components to .ko?
<gnexus>
more about kernel size itself - i need to try some benchmarks to verify what they say
<gnexus>
moving drivers to .ko is what gets the smaller kernel
<gnexus>
i just know i've seen much improvement in "responsiveness" w smaller kernel
<gnexus>
whether that corresponds to actual perfomance gains is yet to see. . .
<mnemoc>
moving random most-probably-unused drivers to .ko is one thing, moving essential drivers used in *all* A10-based devices is another
<gnexus>
mnemoc: quick small easy kernel benchmark?
<gnexus>
no you don't want to move them *all* ;)
<mnemoc>
you were ranting about not been able to move the usb controller to .ko....
<gnexus>
usb does need to be moved
<mnemoc>
I disagree
<gnexus>
np ;)
<gnexus>
the work that it would take would be too much anyway. . .
<gnexus>
one thing that does need to be in defconfig, however, is lzma compression instead of gz
<gnexus>
no drawbacks - much benefit!
<mnemoc>
does uImage support lzma?
* mnemoc
trying
<gnexus>
sure it does
<RaYmAn>
Why would uImage care? iirc it's purely a bootloader format and decompression code is in kernel :)
<mnemoc>
I wasn't sure if the tool that makes the uImage looks inside or not
hark has joined #arm-netbook
ssvb has quit [Ping timeout: 265 seconds]
<mnemoc>
gnexus: beside CONFIG_KERNEL_LZMA, what else?
<mnemoc>
i mean, what other "no drawbacks" and ready to run config changes you suggest?
penguin42 has joined #arm-netbook
<gnexus>
mnemoc: no other changes - no drawbacks
<gnexus>
they must have inherited that gz crap from a VERY old kernel
<rm>
I'd say check what x86_defconfig does
<rm>
if it's not A10-specific, absolutely no reason to deviate
<rm>
e.g. if x86 uses gz, then use gz even if LZMA is "better"
<mnemoc>
the immediate impact is shrinking default uImage from ~4M to 2.8
<mnemoc>
but aligning with common distribution choices is a good idea imo, patches welcomed ;-)
<mnemoc>
more considering providing apt/yum compatible binary repos is one of the goals
zerosheds has quit [Ping timeout: 246 seconds]
<rm>
since this is a kernel it should align to vanilla Linus sources
<rm>
at first I thought "let's copy Debian choices or whatever other distro"
<rm>
but that's not really the best way
<mnemoc>
there is a middle point
<mnemoc>
we are far far away from having anything remotely ready to try mainlining
<mnemoc>
the main use of this tree is people using it drop-in in debian, ubuntu and android systems.... not sure about the adoption in other distros
<mnemoc>
obviusly we won't include all the random patches distros use, but imo defconfig should provide all the basic components to run a real system.. just not going too fancy as we aren't *that* general purpose
<mnemoc>
the 3.4 branch is another story... "users" are sort of happy with 3.0 so 3.4 can aim more at mainlinging and make heavier cleaning
<mnemoc>
iirc debian/ structure includes it's own defconfig, separated from arch/arm/configs/...
<mnemoc>
so taking care of distro-support there lets arch/arm/configs/sun4i_defconfig align better with Linus tree
<mnemoc>
no idea about the rpm world
<hno>
mnemoc, most distributions work fine with current mainline kernel. Patching is only needed in those that insist on running some old kernel.
<hno>
or for hardware not yet mainlined, such as A10..
<mnemoc>
ah, great :)
<mnemoc>
back in 20m
<Turl>
lzma is slow to decompress afaik
graffiti has joined #arm-netbook
<Turl>
mnemoc: beside it's not supposed to be other than 0 or 64, that's why I also suggested moving it to #if inside the code
<Turl>
besides*
<mnemoc>
Turl: the important part is to move mali allocation to a memblock_reserve()
nibb_ has quit [Read error: Connection reset by peer]
<mnemoc>
Turl: playing around the current hack is not an option
<Turl>
well, apparently mali didn't like the fact that linux knew about the ram much :|
<mnemoc>
yes, an the cause needs to be debugged and solved.
<Turl>
did you see the kpanic?
<mnemoc>
yes
<mnemoc>
couldn't continue, but it's because mmu->core is NULL
<mnemoc>
I mean, I couldn't continue debugging. will resume that tonight
<mnemoc>
[ 0.000000] sun4i: core: G2D: disabled
<mnemoc>
[ 0.000000] sun4i: core: LCD: disabled
<mnemoc>
^--- that is also bothering me
<mnemoc>
those memory blocks are not been reserved on my mele
<Turl>
"disabled"?
<Turl>
I disabled g2d on my defconfig as I don't use it
<Turl>
but it doesn't print disabled
<Turl>
it just doesn't print
<mnemoc>
nah, that's a test pr_debug
<mnemoc>
in in the reserve_foo() function
lab5 has quit [Read error: Connection reset by peer]
<mnemoc>
when sys_config tells to not use the thing
<Turl>
sys_config controls that stuff? :P
<mnemoc>
for those, yes
* Turl
prefers to kill them with fire on defconfig
<WarheadsSE>
A100/A1000/A2000 lord knows what else might come on the A10 based line
<Turl>
MeleAX00(0?)
<traeak>
a1000 looks like its toast now
<furan>
hrm. when adding a device to the wiki is it okay to include tips for various hardware modifications to the device or is it strictly device specs
<WarheadsSE>
Thats a hell of a url Turl
<Turl>
WarheadsSE: what about "Mele"
<WarheadsSE>
mnemoc: btw, the 3.0.36 fixed the ntp problems :)
<mnemoc>
uh, I didn't know there were ntp problems
<WarheadsSE>
3.0.8 that I was running from someone would crap on an odd message with ntpd/openntpd trying to set time
<furan>
ok I'll assume it's fine
hipboi has quit [Quit: Leaving]
<WarheadsSE>
furan: if not, it cal always be moved to DEVICE-mods
ka6sox is now known as ka6sox-away
<furan>
where are those?
<mnemoc>
what about sticking the hole mele A100/A1000/A2000 in the page of the first of them?
<mnemoc>
and making redirects from the others to a section about the differences
<WarheadsSE>
Thats fine with me, just trying to sort the proper "landing" page
<mnemoc>
WarheadsSE: good to know our 3.0.36 doesn't have that ugly problem :)
<hackandfab>
I've uploaded an image of the bottom part of the A2000 to the meleA1000 page since it already has the rest of the models
<Turl>
hipboi_: do you know if the nand mbr the kernel uses is a standard mbr?
<mnemoc>
MeleA1X sounds kind of confusing to me because the "A1X" alias is already ised to refere to the socs
<mnemoc>
Turl: doesn't
<furan>
MeleAX
<mnemoc>
very ugly legacy proprietary stuff.... someone should try to get a mtd_sunxi
<rm>
what is wrong with just "Mele"?
<Turl>
mnemoc: I exposed the full nand as a block device but fdisk doesn't like it
<Turl>
rm++
<rm>
also keep in mind that A LOT of stuff is generic between Mele/MK802/MiniX/god knows what else
<mnemoc>
Mele can be a nice landing page listing different devices made by the company
<rm>
I kind of tired seeing "hey get our ubuntu image for Mele A1000"
<rm>
where in fact it would work perfectly not only on A2000 (d'uh!), but also on the MK802
<WarheadsSE>
hackandfab: I must have missed that page.
<hackandfab>
But where is the equivalent recovery button on the mele2000 (I'm reading the boot process at rhombus-tech) and it says there is a specific button on the tablets.
<mnemoc>
WarheadsSE: or it's a permissions thing?
<mnemoc>
hackandfab: not a button, a test point you have to short
<mnemoc>
hackandfab: look for "FEL" in the wiki in RT's wiki
<hackandfab>
ok thx
eFfeM has joined #arm-netbook
<mnemoc>
hackandfab: it would be awesome if you can bring that knowledge to our wiki too ;-)
<traeak>
mele a10 based products :-p
<mnemoc>
we know 3 products using the same A10-based board... that doesn't mean it's the only
<furan>
what do I disassemble to figure out the combos for getting into recovery mode on a device?
<furan>
I cannot for the life of me get this device into recovery boot
<Turl>
furan: you press combos until you get to it :P
<Turl>
usually it's power + volume
<furan>
tried adb, tried different combos
<mnemoc>
WarheadsSE: I just deleted it
<hackandfab>
mnemoc, I'll
<mnemoc>
hackandfab: great :)
<furan>
weird. power + voldown gets me into the same situation where holding the reboot button and plugging in usb does (boots from brom I think)
<furan>
lsusb shows me the same device, anyway
<zenitraM>
furan: that "reboot button" is actually a button to get into FEL, which is what you're seeing.
<zenitraM>
try pressing the other side of the volume button
<furan>
FEL? doesn't it just boot from BROM in the a10?
<furan>
interesting. livesuite doesn't like that mode.
<furan>
so what's FEL?
gnexus has quit [Quit: has left the channel.]
zewelor has quit [*.net *.split]
Gujs has quit [*.net *.split]
domidumont has quit [*.net *.split]
CIA-122 has quit [*.net *.split]
furan has quit [*.net *.split]
<Turl>
ZaEarl: does zatab have a microphone?
<Turl>
nvm answered my own question :)
eFfeM has quit [Ping timeout: 255 seconds]
t0dbld|work has joined #arm-netbook
domidumont has joined #arm-netbook
zewelor has joined #arm-netbook
Gujs has joined #arm-netbook
furan has joined #arm-netbook
CIA-122 has joined #arm-netbook
eFfeM has joined #arm-netbook
<furan>
Turl: devicename_full.mk hardcodes the example tablet device.mk as an include instead of devicename.mk
<furan>
that's why it wasn't picking up proprietary files, I'm pretty sure
<furan>
building now
poocifer has joined #arm-netbook
<poocifer>
anyone have a mele a1000/a2000 with a sata connected?
<mnemoc>
yes
<poocifer>
android or linux?
<mnemoc>
linux
<poocifer>
i'm using android. my sata gets mounted to where only root can write to it.
<poocifer>
i can't figure out how to remount to so an ftp server app can write to it.
<mnemoc>
i would have assumed it used media_rw group for that as with the SD
<poocifer>
it does, but the sd card uses sdcard_rw
<mnemoc>
iirc there is a trick to set the media_rw group to apps asking for permission only to use sdcard_rw
<mnemoc>
adding a line in permissions.xml or something like that
<mnemoc>
but i don't look often inside android systems.... it makes me sick
<poocifer>
that's worth looking into
<poocifer>
it's a ntfs drive and it gets mounted with the default_permissions option
<poocifer>
apparently if I can change that to defer_permissions, I'm golden, but I can't get it to change
<mnemoc>
no clue, too deep into android's jungle for me
<poocifer>
damn. yeah it seems like hardly anyone's actually running android on these things
<rm>
ha
<rm>
true
<rm>
and that's great :)
<mnemoc>
and android + sata + NTFS! is even more fancy
madmalkav has joined #arm-netbook
<poocifer>
should i just reformat it to fat32?
<mnemoc>
that would be easier
<Turl>
poocifer: the stock roms of these things use ntfs-3g from what I saw
<Turl>
and fuse
<Turl>
weird stuff :P
<mnemoc>
but you'll have file size limits
<WarheadsSE>
performance hell
<WarheadsSE>
and partition size limits.
nibb has quit [Read error: Connection reset by peer]
nibb has joined #arm-netbook
<poocifer>
that's doesn't sound ideal...
<rm>
why even run your FTP server on Android
wladislaw has joined #arm-netbook
gimli has joined #arm-netbook
<mnemoc>
why even run an FTP server at all
<mnemoc>
it's so... 90s
<poocifer>
it doesn't have to be ftp, i just want a way to remotely drop shows new shows on the hard drive
<traeak>
ftp is also the most efficient protocol as well
<traeak>
i prefer ftp for local network transfers
<traeak>
but that's usually for 100's of gigabytes
j1nx has joined #arm-netbook
<poocifer>
what if i hooked up the sata drive via usb?
<mnemoc>
not a nice idea
<rm>
why not use the sane way instead
<rm>
and google for DLNA
<rm>
specifically minidlna
<rm>
or whatever kids use to mount SMB shares on android
<rm>
I heard there's even an app for that!
<poocifer>
i set up a shared folder on my win7 pc and successfully mounted it from android
<poocifer>
but I would have to be sitting in front of my mele to get any shows on it
<poocifer>
hence I would like a server so that I can put shows on it remotely
<ManoftheSea>
ftp is efficient?
<ManoftheSea>
how about rsync?
<poocifer>
rsync? on android?
<rm>
poocifer, you're missing the point
<ManoftheSea>
wait, you're downloading a show on android?
<rm>
wtf with "putting new shows on it"
<rm>
just play them from the shared folder, no?
<ManoftheSea>
Why wouldn't you drop the torrent link to the server in the first place?
<poocifer>
rm, I already bought the hard drive for it, so just playing from a shared folder would be waste
<poocifer>
and I don't want the mele to have to rely on the PC at all
<poocifer>
manofthesea, i get my shows through IRC while remotely connected to my arch machine over ssh
<ManoftheSea>
Can we restat the problem? There's an android phone, a mele, and we want to watch shows?
<poocifer>
once i have them d/l'ed, I want to just plop them on the mele and have them ready to go
<poocifer>
starting over...
<ManoftheSea>
Yeah, that's a problem that DLNA solves.
<WarheadsSE>
actually
<WarheadsSE>
he wants to STORE them on the MELE
<ManoftheSea>
Sorry, my posts are slow.
<poocifer>
My mele runs android and I'm going to use it as a media center.
<WarheadsSE>
mele running Android
<ManoftheSea>
The Mele should be a MediaRenderer, the Arch-box a MediaServer, and you can have a ControlPoint in there somewhere.
<ManoftheSea>
But, you want shows to be stored on the android?
<ManoftheSea>
yeah... I know very little about android.
<poocifer>
i can get shows to the mele now, but I have to be a root user
<poocifer>
so adb is the only thing working at the moment
<traeak>
hehe, you'll have to find some app store program that allows you to drop files onto the mele then
<traeak>
or somethign like that
<traeak>
ie: it's not a mele specific problem
<traeak>
but for small shows ethernet whatever should be sufficient
<ManoftheSea>
ethernet should be faster than your display rate.
<ManoftheSea>
Compare: Netflix.
<poocifer>
you're right, it's not a specific mele problem, it's an android problem
<WarheadsSE>
yeup
<poocifer>
thing is, who has a ntfs sata hooked up to android besides meles?
<WarheadsSE>
other androids with sata :p
<poocifer>
good point
<ManoftheSea>
didn't xbmc get something on the mele?
<ManoftheSea>
it's a MediaRenderer
<mnemoc>
using neon
<mnemoc>
i.e. sucks
<poocifer>
there's an early build of xbmc for android in the wild
<ManoftheSea>
ah
<poocifer>
good start, but nowhere near ready for prime time
<ManoftheSea>
use the master ball and catch it.
<ManoftheSea>
(btw, master ball is a regular poke ball wearing wario's thong. CAN'T UNSEE)
<traeak>
xbmc was released for android proper
<poocifer>
the source yes, but no official builds yet
<traeak>
SOCs with sata are pretty rare
<mnemoc>
no cedarx support, no hd
<traeak>
even rarer for the sata to be broken out
<traeak>
mnemoc: so nobody's android builds have cedarx on them ?
<traeak>
or that's just xbmc i gather?
<mnemoc>
traeak: cedarx support currently only works for players embedding allwinner's player
nibb has quit [Read error: Connection reset by peer]
nibb has joined #arm-netbook
<mnemoc>
traeak: fine for me, but forget about xbmc working that way
<mnemoc>
same for vlc
<poocifer>
i gotta run. i might try converting the ftp server to a system app using titanium backup... thanks for the help.
Hexxeh has joined #arm-netbook
<traeak>
mnemoc: although xbmc has access to the cedarx libs...if they work that is
poocifer has left #arm-netbook [#arm-netbook]
<mnemoc>
traeak: afaik gimli's work to support cedarx libs directly in xmbc has nothing to do with xmbc-for-android
<WarheadsSE>
I plan on trying xbmc on my A100 once i get the x11 driver to compile
<WarheadsSE>
seems its being stupid finding drm.h ..
<WarheadsSE>
havent had time to focus on it.
<WarheadsSE>
UMP, done, Mali, done, opengles for mali, done
<mnemoc>
drm.h comes from the kernel headers
<WarheadsSE>
x11, asshat
<WarheadsSE>
Yeah, and it's in the lib path!
<WarheadsSE>
like I said, didnt have time to focus
<WarheadsSE>
meds made me sleepy last night
<WarheadsSE>
[2012-07-17 09:40] ==> Kernel modules are now only read from /usr/lib/modules, all custom
<WarheadsSE>
oops?
<WarheadsSE>
damn left click
Tsvetan has joined #arm-netbook
eFfeM has quit [Quit: Leaving.]
<traeak>
does the a100 have the sata header on the board?
<WarheadsSE>
no
<WarheadsSE>
just need to solder it on, the pads are there
Tsvetan has left #arm-netbook [#arm-netbook]
<traeak>
what a shame, requiring soldering
<mnemoc>
traeak: that saves them $10
<mnemoc>
and for those who can solder and have easy access to the components, it's a good deal
<mnemoc>
(not my case)
<traeak>
saves them 10usd? licensing costs or what?
<traeak>
oh you mean for the entire product
<mnemoc>
A100 is sold $10 cheaper
<traeak>
just leaving on the sata header would be nice
<traeak>
i'm sure that header is even just pennies if that
poocifer has joined #arm-netbook
<specing>
SATA hdr is proly 0.2$
<mnemoc>
considering the price difference between A1k and A2k I'm sure they don't base the prices in cost
gimli has quit [Ping timeout: 248 seconds]
<poocifer>
I figured out how to fix my sata mounting issue with android, if anyone cares
Mazon|M has joined #arm-netbook
<traeak>
.20usd? that much?
<traeak>
hmm
<WarheadsSE>
bascially, it allows tehm to skip a step in mfc
<mnemoc>
poocifer: how?
<poocifer>
First I unmounted /mnt/sata, then remounted it using ntfs-3g, which removed the permission setting
<WarheadsSE>
that times 100K + time saved?
<specing>
0.3 $ on mouser in mass
<mnemoc>
poocifer: so no persistent way?
<specing>
probably 0.1$ in china
<specing>
or even less at 100k
<poocifer>
I used this line specifically "ntfs-3g /dev/block/vold/8:1 /mnt/sata"
<specing>
5 cents?
<mnemoc>
and $1.5+shipping for a normal person? :)
<specing>
ewww ntfs
<poocifer>
I already have a startup script to remount /system RO, so I'll just add this to it.
<specing>
mnemoc: mouser is $40 shipping if you buy <150$
<rm>
if they would mount the header, then could as well wire it up to the back panel
<rm>
as an eSATA socket
<WarheadsSE>
but the A100 really doesnt have the room
<specing>
there must be something else missing on the A100
<mnemoc>
specing: ouch
<traeak>
rm: that would be ideal...but if no competitors are doing that, why shoudl they? who cares about value added? :-p
<rm>
but probably that doesn't fit the 'product vision', marketing etc
<WarheadsSE>
There is nothin different that I have found
<WarheadsSE>
the A1000 has to pay for the USM dock licensing
<rm>
also, has anyone even confirmed that A10's SATA is hotplug-ok?
<WarheadsSE>
^
<poocifer>
Should I be using relatime or noatime with the sata drive?
<furan>
Turl: got mali working. still don't have root ;(
<WarheadsSE>
poocifer: realtime should be ok.
madmalkav has quit [Read error: Connection reset by peer]
<mnemoc>
poocifer: relatime
<specing>
poocifer: first you should stop using ntfs
<WarheadsSE>
^^^
<mnemoc>
+1 specing :)
<specing>
that is why is doesent mount :P
<specing>
it*
<poocifer>
a lot of ntfs hate round these parts
<xenoxaos>
because it's slow on linux
<mnemoc>
poocifer: it's rare to find irc channels where people likes windows stuf... unless it's a "warez" channel
<poocifer>
Relatime for sd and the internal memory as well?
<mnemoc>
yes, everywhere
<rm>
mnemoc, afaik there's #windows :)
<mnemoc>
here on freenode?? uh
<poocifer>
I do use linux a lot, but I run it in a VM, so I figured it'd just be easier to go ntfs.
<specing>
afaik that is the channel unix sysadmins converge to help eachother when they are told to administer that crap
<Turl>
furan: you can echo persist.adb.root=1 >> /system/build.prop too as a temporary measure
<mnemoc>
specing: :D
<poocifer>
of course, i've spent half the day just trying to get ntfs and android to cooperate.
<Turl>
I think that was the right property
<rm>
516 people
<rm>
and it's a ##, naturally
<mnemoc>
poor guys
<specing>
s/people/poor sysadmins/
<rm>
sysadmins are not people? :<
<furan>
ok. I have a driver I want to build and get in there. I know how to modify the .rc file to get it modprobe'd but I am not sure how to get it in the rom image
<furan>
it's the ft5x_ts driver
<furan>
do I need to ask for it in the arch file?
<specing>
rm: sysadmins are strange creatures
<furan>
ls
<DonkeyHotei>
FML FML FML FML FML FML FML FML
<DonkeyHotei>
the seagate freeagent goflex 1.5TB is 14.8mm, won't fit in the mele even as a bare drive
<DonkeyHotei>
now idk what to do with it
<WarheadsSE>
lawl
<xenoxaos>
you could give it to me
<xenoxaos>
:P
<WarheadsSE>
or me :p
* xenoxaos
called dibbs first
<DonkeyHotei>
bah, you're both on the east coast
<xenoxaos>
i'll give you shipping :P
nibb has quit [Read error: Connection reset by peer]
<furan>
turl: any ideas?
<DonkeyHotei>
shipping + what it cost me?
<xenoxaos>
no
<xenoxaos>
:P
nibb has joined #arm-netbook
* xenoxaos
has enough storage atm
<DonkeyHotei>
that doesn't help
<Turl>
furan: enable it on your defconfig
gimli has joined #arm-netbook
<furan>
got it
<poocifer>
donkeyhotei, I thought the mele only supported up to 1 TB
<furan>
Turl: can I just do brunch again after modifying a defconfig or do I have to make clean?
<furan>
trying make mrproper and then brunch
<Turl>
yeah you can
<DonkeyHotei>
poocifer: well, no one told me that
von_fritz has joined #arm-netbook
<furan>
hot brunch is super fast for small changes
<poocifer>
donkey, you have a mele right?
Quarx has quit []
<furan>
what does it mean when it says safe mod ein the bottom left of the screen?
<WarheadsSE>
decent cedarx for *everyting* linux would be nice.
<thefrog>
is CedarX the name of the protocol that supports their hw accel part for the codecs or is it the name of the hw accel part in the A10 chip?
<zenitraM>
i hope they don't discover who actually gave out that service@... email address
<thefrog>
It is on their website
<zenitraM>
ah, nevermind then.
<j1nx>
I mentioned XBMC, because the email communication from within official xbmc email address to allwinner support kind of came to a dead end
<j1nx>
and that service address is already out on the internet
<thefrog>
I mentioned XBMC when I ordered my A100 (A1000)
<furan>
turl: it built the ft5x driver, but didn't put it in the rom - it's only in the obj dir in the device out directory, but not proped to 'system'
<Turl>
are you sure you're using the right defconfig?
Schnabeltier has quit [Ping timeout: 248 seconds]
<Turl>
check the build log
<furan>
yeah it said it built it, and it didn't do that before
<furan>
and it couldn't find that defconfig before i added it
<furan>
all I've done is CONFIG_TOUCHSCREEN_FT5X_TS=y
hipboi has joined #arm-netbook
RaYmAn_ has joined #arm-netbook
j1nx has quit [*.net *.split]
hipboi_ has quit [*.net *.split]
RaYmAn has quit [*.net *.split]
lundman has quit [*.net *.split]
Byan has quit [*.net *.split]
Byan has joined #arm-netbook
RaYmAn_ is now known as RaYmAn
<Turl>
furan: =y makes it built in
<Turl>
=m makes a .ko
j1nx_ has joined #arm-netbook
<furan>
doh
<furan>
does that make a difference?
<Turl>
well if you intend to load a .ko you need =m
<furan>
e.g. if it's built in shouldn't it still work?
<furan>
as I am getting no touch
<Turl>
yeah it should
<Turl>
but these modules on the kernel tree don't always work as advertised :)
<furan>
ok, building module :)
Pere_ has joined #arm-netbook
<WarheadsSE>
theres X11 on arch :)
<WarheadsSE>
damn thing...
<traeak>
arch pulled a nasty with their switch
<WarheadsSE>
ppft, tell me about it
<WarheadsSE>
we got about ? "oops" warning
<WarheadsSE>
yeah, should be kmod and glibc at the same time! (uh..thanks)
<traeak>
hehe
<WarheadsSE>
thankfully upstream is helping us get it all sorted asap, and the build system will crank the shit out of it soon
<traeak>
i feel the pain because my arch systems are some years old
<traeak>
i have /usr on an lvm
<traeak>
on several systems
<WarheadsSE>
Yeah, that upstream move, we didnt see coming
<WarheadsSE>
kmod updated before glibc, .. massive pissed of kernels.
<WarheadsSE>
"no boot!!!1!!111" *le sigh* "mv /lib/modules /usr/lib/modules"
<WarheadsSE>
then they can update, and all is well because the kernel packages have been updated..
<WarheadsSE>
"fun"
<WarheadsSE>
out for a bit
lundman has joined #arm-netbook
thefrog has left #arm-netbook [#arm-netbook]
von_fritz has quit [Quit: vonfritz leaves, don't panic]
<Pere_>
anybody has a mele a1000/2000 working with a joystick? thanks. I've have a ics, but joystick aren't detected
<traeak>
ugh ics
<traeak>
is it supposed to detect joysticks?
<Pere_>
mmm when i pluged in linux mint dmeg say usb conected and i can play, but in mele working android, in terminal ide, dmeg, only says usb conected, but no the input usb
<rz2k>
Pere_: I believe defconfig for mele has joystick framework enabled but no actually joysticks selected to built in kernel. there are bunch of them (common protocol, xbox 360 joystick etc.). you should recompile kernel with desired joystick support.
<traeak>
i do recall plugging a joystick into the mele running linux
<traeak>
that was fine :-p
<traeak>
otherwise your question is a more general android one
<zenitraM>
probably a standard HID joystick
<zenitraM>
other nonstandard joysticks (i.e the original xbox one) aren't HID-compatible and so they require kernel support
<hno>
hackandfab, The correct name for RECV is uboot / FEL (uboot in pinout / schematics, FEL in the rom software that gets activated in the CPU)
<hno>
location of that button in the mele is listed in the wiki.,
<steev>
omfg
<steev>
the nexus 7 is entirely worth it.
<traeak>
you got one?
<steev>
yes
<hno>
the "old" one (rhombus-tech)
mpthompson has quit [Ping timeout: 250 seconds]
<steev>
and if i could have sex with it, i would
<steev>
it's that good
<traeak>
there's probably an audio jack that should work fine for you :-p
<RaYmAn>
awesome
<steev>
haha
<rz2k>
lol
* RaYmAn
is still waiting on his to ship :(
<steev>
RaYmAn: sorry :(
<steev>
i got the 8gb, and no case or anything, just the tablet
<RaYmAn>
mnemoc: more and more devices pushing back adam in the queue ;)
<RaYmAn>
yeah, same
<traeak>
my parents want a 10" though
<traeak>
hmm
<traeak>
it's hard 199 for the nexus?
mpthompson has joined #arm-netbook
<RaYmAn>
EU prices are slightly higher...taxes, VAT and shit has to be included in price here
<steev>
yeah 199 for 8gb
<steev>
USD
<steev>
249 for the 16gb
<RaYmAn>
but still cheap for quad core tegra3 :)
<steev>
i may end up getting the 249 one later
<steev>
and passing this one on to a family member
<traeak>
microsd baby
<RaYmAn>
and high-ppi screen
<RaYmAn>
no microsd :(
<traeak>
wth ?
<RaYmAn>
indeed
<RaYmAn>
It can't safe that much money, so it has to be an explicit decision to not include it
<traeak>
amazon copycat
<RaYmAn>
gnexus doesn't have sd either afaik?
<hno>
Turl, the "MBR" used by Allwinner is not a "DOS MBR". The format is detailed in drivers/block/sun4i_nand/nfd/mbr.h
<hno>
you can access it from u-boot by using partition 0 ("raw" block disk), but not sure how to from Linux.
<Pere_>
traeak, zenitraM, rz2k, in my workin linux mint this is the output
j1nx_ has quit [Quit: Never put off till tomorrow, what you can do the day after tomorrow]
<traeak>
just plugged an axispad into my mele and it was dtected fine :-p
<traeak>
i don't care about ics though
<Pere_>
thanks traeak
<Pere_>
what version do you have?
<hno>
Turl, interesting. Why did I not think about adding a partition that spans the whole disk?
<hno>
probably because what I am really trying to get at is the raw flash where boot0/boot1 resides.
<hno>
which is outside the block emulated layer.
wladislaw has quit [Ping timeout: 240 seconds]
<ZaEarl>
hno, how does boot0/1 relate to boot.axf?
<hno>
ZaEarl, boot0 configured DRAM and loads boot1, boot1 then loads boot.axf from the FAT boot partition.
<hno>
s/ed/es/
<ibot>
hno meant: ZaEarl, boot0 configures DRAM and loads boot1, boot1 then loads boot.axf from the FAT boot partition.
<hno>
For u-boot NAND SPL I need to change boot0 to u-boot SPL, and boot1 to u-boot.
<ZaEarl>
so boot.axf is the first thing bootable from a normal partition.
<hno>
plus somehow strip down the A10 nand driver so it fits in SPL.
<hno>
ZaEarl, yes.
<hno>
both boot0 and boot1 lives outside the virtual block device layer.
<ZaEarl>
is there some source for boot0/1/afx?
<hno>
no
<hno>
well, allwinner naturally have some source, but it's not open not even to allwinner customers.
<hno>
seems historically boot.axf source was open to customers, but no longer.
<hno>
i.e. in some earlier chip generation.
<ZaEarl>
I'm guessing FEL mode is before boot.axf.
<hno>
Long before. It replaces boot0
<hno>
Or actually it replaces the code which loads boot0.
<ZaEarl>
so if you trash boot0/1, you can still recover using LiveSuit?
<hno>
yes.
<ZaEarl>
nice
nibb has quit [Read error: Connection reset by peer]
<hno>
or by using a phoenixcard.
nibb has joined #arm-netbook
<Turl>
hno: so pretty much the only way to brick an a10 is by burning the cpu right? :P
<hno>
Turl, yes, only way to brick an A10 device is to actually crash the hardware.
<hno>
and as long as the CPU & USB OTG port works then FEL mode works, even if both NAND and DRAM is broken.
<hno>
not that you would be able to do very much with the device is DRAM is broken. But there is about 256KB internal SRAM so should be possible to use a almost raw A10 CPU as a fancy USB connected GPIO controller even without any DRAM.
<hno>
not all of that is immediately addressible however, most needs configuration to give the ARM CPU access to it.
<WarheadsSE>
:) xfce running.. odd behavior from display ..
<Turl>
hno: otg port?
djdonkey has joined #arm-netbook
<ZaEarl>
I use LiveSuit with the non-host port.
<djdonkey>
hi all, now I'm sure you get a 1000 of these a day, but i'm trying to rev-engineer my own android ICS image onto a known working A10 tablet. I use ICS 4.0.4 and Kernel 3.0.8. It actually boots up, but I can't for the life of me get the goodix touchscreen driver to be recognised!?
<djdonkey>
has anyone hacked into this process? the known working image runs 4.0.3 and kernel 3.0.8, and loads goodix_touch.ko
* markvandenborre
has two cameras running from a Mele A2000; thanks all!
<djdonkey>
i know i'm missing something simple... my fex is identical i believe too
<ZaEarl>
djdonkey, are you compiling Android from source?
<djdonkey>
hi, ZaEarl and yes
<djdonkey>
i have access to official allwinner
<djdonkey>
and using that
<ZaEarl>
there are a lot of goodix models, you need to have the source to one that matches your hardware
<djdonkey>
gt801
<djdonkey>
got it
<djdonkey>
get this
<djdonkey>
i even build the gt801 module from my github
<djdonkey>
and i copy it over to the known working A10 tablet
<djdonkey>
over writing its goodix_touch.ko
<djdonkey>
i even put debug statements in to make sure it is "my built" module
<djdonkey>
and it works!
<djdonkey>
which is why i think in my build i'm doing something fundamentally wrong
<djdonkey>
its like i have the fex wrong, specified the wrong pins...
<ZaEarl>
yes, it does sound like something else is wrong
<djdonkey>
i keep coming back to the fex
<djdonkey>
maybe i dont have it copied 100%
<djdonkey>
from known working tablet
<ZaEarl>
how are you installing Android, recovery or LiveSuit?
<djdonkey>
LiveSuit
<ZaEarl>
You'll want to make sure the [ctp_para] section is correct.
<djdonkey>
ZaEarl
<djdonkey>
i've gone blind trying to find the differences
<djdonkey>
i believe it to be EXACT
<djdonkey>
i mean goodix chip is hardwired to 0x55 bus addr
<djdonkey>
i know the tablet is wired to twi int 2
<djdonkey>
and id is 1
<ZaEarl>
you can just grab a known-working script.bin and drop it into place after you install to verify if that is the problem..
<djdonkey>
hmmmm
<djdonkey>
now thats something i haven't done
<djdonkey>
is it a dumb question to ask what is script.bin
<djdonkey>
?
<ZaEarl>
the binary form of the fex
<djdonkey>
ZaEarl, champion
<djdonkey>
i will try this
<djdonkey>
something new to try
<djdonkey>
pulling my hair out!
<ZaEarl>
script.bin is in nanda, which is easily mounted as a vfat partition
<djdonkey>
why didn't i just join this channel 3 days ago
<djdonkey>
and sit here waiting for wits-tech to answer my QQ requests.... :(
<djdonkey>
we have pulled the script.bin from the known working tablet, and decoded
<djdonkey>
but never copied back into our image
<djdonkey>
in binary...
<djdonkey>
so maybe we left something off...
<djdonkey>
i suppose this way we'll be 100% sure
<djdonkey>
if this doesn't work, what else?
<djdonkey>
i'm thinking that perhaps the gt801 is not even awake? like no power or something?
<djdonkey>
and this would lead me to look at the axp902?
<hno>
Turl, A10 USB0 controller is an OTG controller, capable of both HOST and DEVICE modes.
<djdonkey>
ZaEarl, just booted up with known working script.bin, still same issue
<hno>
USB1 & USB2 is HOST-only controllers and is now usable for FEL booting.
<hno>
ZaEarl, there is no non-host ports on A10.
<hno>
usb
<ZaEarl>
djdonkey, well, you've eliminated one potential problem area.
<Turl>
hno: my tablet has a 'host' (OTG) and then the typical 'client' port
<djdonkey>
ok for this i am grateful
<djdonkey>
ideas for next step...
<ZaEarl>
throw some more logging into your module and see where it's failing
<djdonkey>
its right at the i2c probing
<hno>
Turl, The OTG port is usually a mini-USB connetor.
<hno>
not sure what you mean by "client" port.
<djdonkey>
sending a simple i2c device probe on twi 2 bus addr 0x55
<Turl>
hno: they're both miniusb
<Turl>
hno: one port to connect to pc only
<Turl>
hno: and another port to connect client devices (eg a keyboard or memory stick)
<ZaEarl>
hno, one usb port can power keyboard/mouse, the other port can handle LiveSuit.
<ZaEarl>
charging the tablet only works on the same port LiveSuit works on.
<hno>
The OTG port can be wired to support both charging and powering of devices.
<ZaEarl>
I've tested a lot of tablets, never found any that were wired that way.
<ZaEarl>
although I suppose it might just need some script.bin tweaks to make work.
zerosheds has joined #arm-netbook
<djdonkey>
ZaEarl
<djdonkey>
will i upset ppl if i cut and paste logs in here?
<djdonkey>
<6>[ 6.320037] Warnning: I2C connection might be something wrong!
<Turl>
yeah probably not the driver for your device
zerosheds has quit [Read error: Connection reset by peer]
<djdonkey>
so heres the funny thing
<djdonkey>
in our source we build goodix_touch.ko
<djdonkey>
and we side load this into the known working tablet that we are trying to rev-engineer
<djdonkey>
and our module works!
<djdonkey>
we even put debugs in our module to be 100% sure that it is our built module
<djdonkey>
so doesn't this mean our code for the module is correct?
<Turl>
the chinese never build two batches the same, ZaEarl can tell you that
<Turl>
your other tablet probably has some other goodix model or something
<djdonkey>
but we use the same image on both of them
<djdonkey>
and both work with our driver
<djdonkey>
but not our AOSP
<djdonkey>
build
<Turl>
so you're saying
<Turl>
you build a kernel+module
<Turl>
put it on both tablets replacing the stock kernel and modules
<Turl>
and touch works on both?
<djdonkey>
100% correct
<Turl>
then where's the issue?
<djdonkey>
hang on
<djdonkey>
not 100% correct
<djdonkey>
:)
<djdonkey>
we have two tablets
<djdonkey>
both built by same manufacturer
<djdonkey>
claim to be the 'same' HW
<Turl>
yeah, 'claim' is keyword there
<djdonkey>
(you made me now aware)
<djdonkey>
but, built at the same time ok
<djdonkey>
both came with an image 100% working
<djdonkey>
their kernel, their module
<Turl>
same image?
nibb has quit [Read error: Connection reset by peer]
nibb has joined #arm-netbook
<djdonkey>
yes same image, as they gave us the updated ICS to run on them
<djdonkey>
so we used LiveSuit to install same image on both
<djdonkey>
now, we have gone to wits-tech and bought the dev board in attempt to rev-engineer
<djdonkey>
and run OUR AOSP onto these tablets
<djdonkey>
this is where we have come stuck
<djdonkey>
because in our environment we cant get gt801 to be recognised
<djdonkey>
so then we build our own goodix_touch.ko module
<Turl>
but are you sure both tablets have a gt801?
<djdonkey>
and side load this onto the known working image
<Turl>
the stock images load several modules
<djdonkey>
yes, both hav gt801's
<Turl>
one might be using gt801 and the other some other
fullofSheds has joined #arm-netbook
fullofSheds has joined #arm-netbook
<djdonkey>
yes... both have same lsmod results
<Turl>
djdonkey: do this to confirm
<Turl>
load the working image on both
<Turl>
then rmmod the same module on both
<Turl>
then confirm or deny both lost touch sensitivity
<furan>
hrm. apparently touch is working but the screen isn't updating. every time I touch the screen logcat spits out "surface::lock failed, already locked"
<furan>
and mail got some unknown symbols when it loaded, such as ump_dd_handle_create_from_secure_id
<djdonkey>
Turl, on it, will report back
<djdonkey>
i appreciate anything you guys can give
<Turl>
furan: maybe your mali libs are too old
<Turl>
furan: let me zip mine up
<furan>
thanks
<hno>
furan, that symbol is in the mali UMP driver