<Superpelican>
hmm I just ran the arm-linux-gnueabihf-gcc command myself and got "The program 'arm-linux-gnueabihf-gcc' is currently not installed. You can install it by typing: sudo apt-get install gcc-arm-linux-gnueabihf"
<Superpelican>
so now installing that
<Superpelican>
rm: apparently that gcc-4.8-arm-linux-gnueabihf package wasn't enough, apt-get now reports: "The following NEW packages will be installed: cpp-arm-linux-gnueabihf gcc-4.8-multilib-arm-linux-gnueabihf gcc-arm-linux-gnueabihf libc6-armel-armhf-cross libc6-armel-cross libc6-dev-armel-armhf-cross libc6-dev-armel-cross libsfasan0-armhf-cross libsfatomic1-armhf-cross libsfgcc-4.8-dev-armhf-cross libsfgcc1-armhf-cross libsfgom
<Superpelican>
rm: It's now making make'ing
<Superpelican>
rm: Looks like it just finished with success
buZz has quit [Ping timeout: 258 seconds]
buZz has joined #linux-sunxi
buZz is now known as Guest7680
leviathanch2 has quit [Ping timeout: 245 seconds]
<aep>
so i got the mali blob thing to load into fbturbo, but its still dog slow and still no GL
<aep>
i wonder what else to do :/
<TheSeven>
what's the state on MTD support and uboot SPL these days?
<TheSeven>
is that usable to boot linux from nand already?
nove has joined #linux-sunxi
Xaros has quit [Quit: Zwi se logou mou!!!]
<torindel>
aep: keep in mind that those are gles not gl libs
<Nazcafan>
kudos to the free electons guys for creating the linux cross reference
<ssvb>
aep: and check which shared libraries it depends on
<aep>
ssvb: not sure what to check there. the libEGL it uses is libMali.so
<aep>
same for libGLESv2
<megal0maniac>
"if an Android image is installed on NAND Flash and a Linux Image is installed on SD card, once you boot into Linux system from SD card, you will not be able to boot into Android system unless you re-install the image"
<ssvb>
aep: check if it is not actually loading mesa shared libraries
<aep>
nope, pretty sure
<megal0maniac>
Is this a known issue on A20 based devices?
<torindel>
aep: if it uses libmali for everything check that you have /dev/mali with correct permisions
<ssvb>
aep: so the ldd log looks sane, that's good
<aep>
i'm root
<aep>
whut, it works after a reboot
<torindel>
megal0maniac: that means your android/linux image provider fucked something up ^^
<megal0maniac>
torindel: Apparently (because I didn't test this personally, and the internet is full of lies,) this affects many (presumably all) A20 boards with many different images
<megal0maniac>
My assumption is that Linux writes to flash for some reason
<megal0maniac>
I don't know what, or why, but I've had the same problem and I don't see how Android could be the culprit
<ssvb>
aep: ok, that's good :)
<aep>
yeah, except still mplayer is too slow
<aep>
i guess i need a player that does gles
<ssvb>
aep: no, mplayer does not need mali
<aep>
well, sure, it can use software scaling, but thats whats too slow
<ssvb>
aep: are you trying something like 1080p videos?
<aep>
yeah, with rotation
<aep>
the decoding part is fine, the rotation is the issue
<ssvb>
rotation?
<aep>
the screen is rotated 90 degree
<megal0maniac>
Although if Linux is installed on NAND then it's fine. So I don't know...
<aep>
software rendering is too slow to do that
paulk-collins has joined #linux-sunxi
<aep>
hence i was hoping for it to use gl(es) to do that, but -vo gl uses gl, not gles
<ssvb>
aep: by using Option "Rotate" in xorg.conf?
<aep>
yeah
<nove>
megal0maniac: i heard something about, if the nand driver cant understand the format of the nand, it will do unconditional format
<ssvb>
aep: yes, this is absolutely horrible, highly non-recommended to use
<torindel>
are gles => gl glue libs worth it? or uber slow anyway?
* torindel
adds it to todo list to test
<megal0maniac>
nove: That seems silly given that an SD image shouldn't need to touch the NAND
<ssvb>
aep: none of the accelerated features work, and you get an extra software rotation step and and extra intermediate memory buffer
<aep>
ssvb: yah, i noticed that it only works for half the X stack anyway :(
<nove>
megal0maniac: the silly in the nand driver from allwinner
<ssvb>
aep: why do you need rotation?
<aep>
because the screen is physically rotated
<ssvb>
aep: do you have some unusual hardware?
<aep>
nah, its a regular lcd connected with lvds
<aep>
but, mounted 90 degree rotated
<ssvb>
"you are holding it wrong" :)
<aep>
:D
<nove>
megal0maniac: the format of the nand changed in new androids, but the linux-sunxi nand driver i think still uses the old one
<nove>
megal0maniac: but there others that know more than i, about this
<torindel>
aep: get lvds connector that does hardware rotation and you wont have problem xd
<ssvb>
aep: are you using it as a regular linux desktop, or want to run some specialized software on it?
<megal0maniac>
I see. That's enough to get the ball rolling though. Thanks :)
<ssvb>
aep: like, for example, just a media player?
<megal0maniac>
nove: libnand is a binary blob?
<aep>
torindel: err, i can just turn around the lvds signal? i thought its hardcoded into something which way is row and which is column
<aep>
ssvb: a browser :(
<aep>
ssvb: which means no gl on fb
<torindel>
aep: nope
<aep>
torindel: is that a no i cannot rotate it or yes i can? :D
<torindel>
no you cannot rotate signals
<aep>
right
<aep>
i tried moving around some pins and it just stopped working
<ssvb>
aep: I don't think that the display controller supports rotation (or maybe I'm wrong?), but g2d does
<ssvb>
aep: just nobody has ever asked for accelerating rotation :)
<aep>
:D
<aep>
i guess the best approach would be to put everything i want on fb with gles
<aep>
instead of using X
<nove>
megal0maniac: the old we have source, but the newer still only a blob
<aep>
thats the kind of day you wish wayland would actually work :(
<megal0maniac>
nove: One day we will have mtd :)
<torindel>
nove: we could just remove that silly format thing from the opensource one, if what you written is true
<ssvb>
aep: how does wayland help with anything?
<ssvb>
aep: gles can also do rotation in X
<aep>
oh neat
<aep>
so i need to teach mplayer to do gles?
<aep>
or use some other player
<aep>
for the browser, i'll think about it later
<aep>
just disables rotation, and video plays butter smooth obviously :/
<ssvb>
aep: mplayer supports gles, but needs to be compiled in a special way
<nove>
torindel: yes, but i am only speaking what i read, never used nand
<nove>
aep: why not instead rotate the video, or the web content
<torindel>
nove: i dont have that problem since i dont use new android + linux ;]
<aep>
nove: the first would be easy, the second is kinda hard :D
<aep>
whut it works with -vo gl!
<aep>
magic
FreezingCold has quit [Ping timeout: 258 seconds]
<megal0maniac>
nove: Any idea who would know more? It doesn't seem to be documented anywhere
<megal0maniac>
libv_ seems to know at least something about everything :P
<megal0maniac>
And everything about some things
geecko has joined #linux-sunxi
linkmauve1 has left #linux-sunxi ["Disconnected: Replaced by new connection"]
linkmauve1 has joined #linux-sunxi
<hramrach>
wingrime: ?
<nove>
megal0maniac: maybe searching in logs for nand will give you some idea
<megal0maniac>
Hmmm... Good idea
nabblet has joined #linux-sunxi
megal0maniac is now known as megal0maniac_afk
<wingrime>
hramrach: maybe cedar support 10bit
paulk-collins has quit [Quit: Ex-Chat]
<nove>
wingrime: my guess, that part is from the bitstream parser, not related to hardware
<wingrime>
nove: maybe
<wingrime>
nove: I added rmvb
<wingrime>
nove: but sad there is no encoder regs
<nove>
wingrime: yes, to find more about isp
<wingrime>
nove: but ARGB seems is alpha blender that we are looked for
<wingrime>
nove: are you plan add new regs to parser
leviathanch2 has joined #linux-sunxi
<wingrime>
nove: I hope we will deal with interlanced
bertrik has joined #linux-sunxi
bertrik has quit [Changing host]
bertrik has joined #linux-sunxi
<nove>
wingrime: but is a bit odd, i wouldn't be surprise if the A of ARGB, is for Anaglyph
<wingrime>
nove: I see "alpha" in tags
<wingrime>
in bitfields
<wingrime>
nove: also seems 1633 can also yv12_output_enable
<wingrime>
nove: and interesting news , we have new F2C700 soc
<wingrime>
nove: cortex a8
<nove>
wingrime: how it can be used to blend to a decoded video frame? this looks a separated block that is used alone
<wingrime>
nove: so, who do YUV to RGB?
leviathanch2 has quit [Ping timeout: 240 seconds]
<wingrime>
nove: can it be codec->ANAGLYPH->ARGB-> rgb output?
<wingrime>
nove: for anaglyph you should shift R G B
<wingrime>
nove: and blend
<nove>
wingrime: too weird, maybe this are vestiges of old hardware, this ARGB registers have "not used in 1633."
<wingrime>
nove: 1633 have no MAF too
<wingrime>
nove: a23 is chap IC
<wingrime>
nove: maybe a13 have no it too
<nove>
wingrime: A31 and A23, maybe is like A10/A20, same video engine
<wingrime>
nove: 000: read\/bule, 001: red\/green, 010: red cyan, 011: color, 100: half color, 101: optimized, 110: yello\/blue, are presend
Web-aptosid876_ has joined #linux-sunxi
<wingrime>
nove: but I not see simple blending
<wingrime>
nove: can you trace andorid ?
<nove>
wingrime: that was what i have been thinking
<Superpelican>
libv: I know
<Superpelican>
I've seen it
<Superpelican>
I should contribute to that page
<Superpelican>
I think I have some useful information
<Superpelican>
for example the original script.bin
<Superpelican>
and I have also converted it to a .fex
<Superpelican>
so it could be added sunxi-boards
shineworld has joined #linux-sunxi
<Superpelican>
not an expert though
<nove>
wingrime: but for sure will be a pain to do so
<wingrime>
nove: you just need tracer build with bionc, and mediaserver should be warped around it
<Superpelican>
libv: BTW you aren't the developer of Lima? ;)
<shineworld>
How to remove some defaults in uboot so my root= in uEnv.txt is the only ? during boot I've saw : [ 0.000000] Kernel command line: console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait loglevel=8 panic=10 root=/dev/mmcblk0p3 loglevel=8 rootwait console=ttyS0,115200 rw init=/init mac_addr=E2:40:02:93:5D:59
<wingrime>
nove: 00: source mode; 01: top\/down or over\/under; 10: line by line; 11: separate
<nove>
wingrime: or we just find out how the binaries api works, using the android sources
<wingrime>
nove: 01 works
<nove>
wingrime: anyway i can't look at it now, maybe in next months
<wingrime>
nove: but 10,11 not works
<wingrime>
CEDARV_3D_MODE_DOUBLE_STREAM
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
<wingrime>
for double stream video like MVC and MJPEG.
xinj has quit [Ping timeout: 258 seconds]
geecko has joined #linux-sunxi
<Superpelican>
I can now see WiFi networks with iwlist wlan0 scan, but trying to connect with wpa_supplicant results in "Failed to initialize driver interface \n Segmentation Fault" with -D nl80211
nedko has quit [Quit: kernel panic]
nedko has joined #linux-sunxi
nedko has quit [Changing host]
nedko has joined #linux-sunxi
<Superpelican>
when omitting -D or using -D wext I get "iotctl[SIOCSIWAP]: Operation not permitted \n ioctl[SIOCSIWENCODEEXT]: Invalid argument"
<nove>
wingrime: but this is strange, in a encoder demo they have done a watermark that is blending by software, with this and not using the hardware
<Superpelican>
going to plugin in the tablet and reboot to see if it helps
<Superpelican>
ok plugging in and rebooting did not work
shineworld has quit [Quit: Leaving]
wingrime has quit [Quit: Lost terminal]
nabblet has joined #linux-sunxi
<jemk>
wingrime, nove: I could not find a way to use anaglyph as simple yuv to rgb converter, it always changes colors and size
<jemk>
if you want to play yourself, wiki is updated with anaglyph register
<nove>
jemk: nice, you made it to work?
<jemk>
nove: I changed my old mpeg-test a little bit and played around with the control register and it produced funny coloured pictures, if you want to call that work
<nove>
jemk: iok
leviathanch2 has joined #linux-sunxi
geecko has quit [Remote host closed the connection]
leviathanch2 has quit [Ping timeout: 245 seconds]
<libv>
Superpelican: by all means, add to that page
<libv>
and a proper .fex file to match the wiki page is really what we need there
<libv>
i couldn't make out which was which from looking at the wiki and then looking at the .fex file
<Superpelican>
libv: Ok, then I'll certainly add it to the wiki, BTW do you have any idea what might be causing those error messages?
<wens>
thinking about buying an A31s tv box to play with
<ssvb>
jemk: hopefully I'll release a usable dram_trp3 tuner tool by tomorrow's evening :) it seems to be capable of tweaking the phases of individual lanes
[7] has quit [Ping timeout: 245 seconds]
<ssvb>
jemk: just testing is really slow, maybe I need to implement a simulator of 'reasonable' memtester results, and plug it in
TheSeven has joined #linux-sunxi
wingrime has joined #linux-sunxi
<wingrime>
jemk: are bits from anglyph are from traces?
<jemk>
ssvb: nice, i will try it. It shouldn't be a big problem if it is slow, it should have to be run only once per board type
<jemk>
wingrime: no, from trail and error
<wingrime>
jemk: how you visualise results?
<jemk>
wingrime: disp scaler can use the three plane rgb
<Superpelican>
then I'll just manually copy over the firmware and hope for the best ;)
<rm>
also what is wrong with just installing the deb?
<rm>
or do you not use debian?
deasy has joined #linux-sunxi
Skaag_ has quit [Ping timeout: 240 seconds]
<Superpelican>
rm: I'm using Linaro/Ubuntu
<Superpelican>
but with a custom kernel
<Superpelican>
just an uImage
<rm>
unpacking will work, but also the deb could work
<Superpelican>
well I copied over all the files
<Superpelican>
like the debian page suggested
<Superpelican>
but it didn't work
<Superpelican>
ran depmod -a to be sure
<Superpelican>
and rebooted
<Superpelican>
but still the same errors
<torindel>
any gcc guru out there? need to know if they fixed optimaling with -mfpu=neon on non-neon aware things (non-neon aware programs compiled -mfpu=vfpv3 where faster), and how goes neon-vfpv4 with that
<Superpelican>
getting a bit desperate :(
<torindel>
Superpelican: check dmesg?
<torindel>
kernel usually says which file and where it cannot found for firmware
<torindel>
and ps. you can make firmware_install from your kernel source
<Superpelican>
torindel: dmesg reports that the emac driver is disabled, could this cause problems?
<torindel>
Superpelican: probably yes, esp if youre trying to use it ;]
<Superpelican>
torindel: And would wpa_supplicant use it when trying to connect to a WPA2-PSK WiFi network?
<Superpelican>
Also what is very strange is that dmesg reports "ADDRCONF (NETDEV_CHANGE): wlan0: link becomes ready"
<Superpelican>
but I after each boot I have to manually activate wlan0 with ifconfig wlan0 up
<Superpelican>
torindel: Ok I've now read the whole dmesg log, nothing about firmware or the rtl8192cu in it
MSameer has quit [Excess Flood]
MSameer has joined #linux-sunxi
<torindel>
Superpelican: emac driver is for 10/100 cable realtec
<Superpelican>
so it isn't needed for WiFi?
Skaag has joined #linux-sunxi
<torindel>
as for "ADDRCONF (NETDEV_CHANGE): wlan0: link becomes ready" thats normal and should show up when wpa_supplicant establishes connection
<Superpelican>
well then it's very strange that I have to first manually activate wlan0
<torindel>
as for your manuall activation just setup your network correctly maybe? xd
<Superpelican>
torindel: What do you mean?
<torindel>
Superpelican: dunno what distribution you use but correctly configured network would bring up interface by itself, run wpa_supplicant, get ip from dhcp/whatever by itself
<torindel>
without manual intervention
<Superpelican>
torindel: using Linaro (Ubuntu Server)
<torindel>
Superpelican: gentoo here ^^
<Superpelican>
torindel: trying to convert an unused tablet to an Owncloud server :)
leviathanch has quit [Read error: Connection reset by peer]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
xavia has joined #linux-sunxi
<Dodger78>
is it possible to kexec kernels , that need a uboot environment ? im trying to test a new kernel before installing it
<Superpelican>
by editing the /etc/network/interfaces
<Superpelican>
& a reboot
leviathanch has joined #linux-sunxi
nove has quit [Quit: nove]
Dodger78 has left #linux-sunxi [#linux-sunxi]
deasy has joined #linux-sunxi
leviathanch has quit [Remote host closed the connection]
leviathanch has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
Superpelican has quit [Ping timeout: 240 seconds]
jemk has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
d3vr has joined #linux-sunxi
<d3vr>
Hello guys, please help me am having trouble trying to root my A23 Q8H tablet... It seems that ADB needs to be enabled but I couldn't figure that out... (http://linux-sunxi.org/Ippo_q8h)
d3vr has quit [Ping timeout: 258 seconds]
Gerwin_J has quit [Quit: Gerwin_J]
paulk-aldrin has joined #linux-sunxi
netlynx has quit [Ping timeout: 240 seconds]
Quarx has quit [Read error: Connection reset by peer]
<torindel>
getting back to my earlier question: any gcc guru out there? need to know if they fixed optimaling with -mfpu=neon on non-neon aware things (non-neon aware programs compiled -mfpu=vfpv3 where faster), and how goes neon-vfpv4 with that