[WIP] AOSP tegra 3 port

[GizmoTheGreen]

why all this messing around to "port" cyanogenmod couldn't you compile from source?
I lack the skill and knowhow myself but it sure would be alot cleaner. and easier to change once it's set up.

 

[professorpoptart]

At this point maybe, the original idea though was, that since the tf700 was so close to the ouya hardware wise, simply changing the updater_script to the point to the correct blocks and copying over a few files for hardware should get us 99% of the way... or at least to a bootable state.

Really the trick is going to be in getting boot.blob right to be able to boot CM. Even if I build the rom... meaning system... from source, I will need to have this research into the ramdisk to boot to it.

Unless anyone has any other ideas! Lol

Sent from my SAMSUNG-SGH-I337 using xda app-developers app

 

[lartomar2002]

Looks like you have been getting some press lately.

 

[professorpoptart]

Looks like you have been getting some press lately.

?, haven't seen anything...

 

[lartomar2002]

Hups, it was sonofskywalker3, my bad:cyclops:

 

[professorpoptart]

Hups, it was sonofskywalker3, my bad:cyclops:

ok...

 

[professorpoptart]

Ok! My dev virtual machine is back up! setup the new one with 2.2TB of storage, that should give me enough to run a few builds from source, if I need to... in the future.

So FYI that 30gb recommended by the kitchen in link on the OP, that really is a minimum. Give it more...

...and snapshot has been made. :good:

 

[sonofskywalker3]

if you decide to do a source build, i'll be glad to share what i've done with the CM10 source, not that it works...

 

[professorpoptart]

Just wanted to take a sec and say how much I love http://www.diffchecker.com/diff

ok got a new AOSP port i did using this tut, doing a backup and flashing shortly...
http://forum.xda-developers.com/showthread.php?t=1598713

should have the pieces needed;
need a jb(4.1.2) rom made for the device (the OTA)
need a jb(4.1.2) rom that I want to port
if going from different manufacturer must have same processor (both tegra 3)

issue with updater_script permissions, getting status 7
...now syslink...

..sure do spend a lot of time uploading this zip to the ouya...

 

[lartomar2002]

Would it help if I were to send you the build prop from my Asus TF700?

 

[professorpoptart]

So tried a new packaged AOSP rom + boot.img, it didn't boot but didn't go into bootloader either, and the monitor had an input signal too. I'll take that as progress. The better news is that stock ouya will still boot with my boot.img! So hopefully that means its more of a rom issue then boot.img. I'll post my boot.img in the op in case anyone wants to test with it, unpack it or do dirty things to it.

Just to test I tried one of the old AOSP builds that got stuck on the boot animation with the new boot.img but same deal... stuck on animation.

tomorrow(edit, I mean the next day. Which is now tomorrow...) the bus pirate should be here and I will either figure it out or end up breaking the ouya!.... or just not figuring it out...

Updated the first post with all the new things

 

[professorpoptart]

Would it help if I were to send you the build prop from my Asus TF700?

Na, from what I understand its more important to use the build.prop from the port rom... well a merger of port and base build.props

Thanks for the offer though.

Sent from my SAMSUNG-SGH-I337 using xda app-developers app

 

[professorpoptart]

OK! I got my bus pirate hooked up, minicom installed and streaming kernel console to the terminal!

First thing I noticed was no wires in the box (I mean bag). As I had no breadboard jumper M/F wire or an 10 pin IDC serial ribbon, I used plain old spooled wire I had laying around. I cut 2 pieces about 6-8 inches long and threaded the wire through the TX and ground UART pinouts on the OUYA. It just so happens the wire is the "perfect" diameter to fit snugly in place in the pinouts. I then threaded the wire through the top ventilation slits in the top of the OUYA so I can close the top and have the 2 wire popping out the top. I tied a knot in the TX one to identify it later on, as its just plain black wire. I then twisted the other ends into coils and slid them onto the ground and MISO pins on the buspirate.

on the terminal run sudo minicom then enter the menu...
HiZ>m
then options;
3
9
1
1
1
2 (edit: ment 1)
at UART> type "(3)"
y

turn on the OUYA and the kernel debug info will start streaming in!

to "hang up" on UART
"crtl+a" then "h"

And now for some kernel debugging!

bought the bus pirate from amazon.
http://www.amazon.com/SparkFun-Bus-...TF8&qid=1370473406&sr=8-1&keywords=bus+pirate

used this to figure out what goes where.
http://dangerousprototypes.com/bus-pirate-manual/bus-pirate-uart-guide/ (this one for the 10 pin IDC to UART pinout name conversion)
http://dangerousprototypes.com/docs/Bus_Pirate_I/O_Pin_Descriptions (this one that actual shows where the pins are)

used this to setup minicom for the buspirate.
http://jumptuck.com/2010/01/20/using-the-bus-pirate-with-ubuntu/

watched this video to actually run minicom.
https://www.youtube.com/watch?v=UUiEzv_LQr0&feature=youtube_gdata_player

 

[sonofskywalker3]

Thats awesome! Glad you're making progress. I might as well have been beating my head on the wall for all the progress i've made, and i'd probably have less of a headache.

 

[mburris]

OK! I got my bus pirate hooked up, minicom installed and streaming kernel console to the terminal!

Nice going. I had the exact same situation and challenges when I was doing the debugging on a Gbox.

The fact that you were able to get it all setup correctly should give everyone a nice dosage of hope.

Post the log of where it is failing, if you don't mind.

---------- Post added at 08:33 PM ---------- Previous post was at 08:27 PM ----------

Just wanted to take a sec and say how much I love http://www.diffchecker.com/diff

On Ubuntu I like Meld http://meldmerge.org/ for diffing a large tree of files (and their content) against another.

 

[professorpoptart]

Post the log of where it is failing, if you don't mind.

Will do, early day tomorrow so I will have to wait untill after work tomorrow to post the full (rom+boot.img) update.zip along with a paste bin log of the uart console for review/input.

*Rambo voice*
Google play store... I'm coming for you...

Sent from my SAMSUNG-SGH-I337 using xda app-developers app

 

[professorpoptart]

On Ubuntu I like Meld http://meldmerge.org/ for diffing a large tree of files (and their content) against another.

Now thats cool!

Sent from my SAMSUNG-SGH-I337 using xda app-developers app

 

[professorpoptart]

!!! As With anything posted anywhere in this thread, these may destroy your device! They are for research and development only!!!

So here is my build from a couple days ago with a merged ramdisk and system from the liquidsmooth zip and 264 OTA zip. Basically the issue with this build is that it gets stuck on the boot animation.

after flashing the LS_OUYA.zip, flash the FixPerms.zip...(bottom of post)
link to LS_OUYA.zip http://www.mediafire.com/download/psmsv1i2qq9muic/LS_OUYA.zip

Here is the link to the paste bin for the full kernel console output via UART.
http://pastebin.com/yFKPsMea

The rom had issues setting permissions with its updater_script so I made a seperate zip just for the perms that need to be run after the rom is installed. Then it will boot up to the boot animation. It then loops on the below segment. With different a "untracked pid xxxx exited" on each loop.

[ 142.123055] nvavp nvavp: using SMMU at ff00000 to load AVP kernel
[ 142.129478] nvavp nvavp: AVP os at vaddr=f0e00000 paddr=ff00000 reset_addr=04
[ 142.144920] nvavp nvavp: using SMMU at ff00000 to load AVP kernel
[ 142.151303] nvavp nvavp: AVP os at vaddr=f0e00000 paddr=ff00000 reset_addr=04
[ 146.960679] TF : INFO 00000063 02:55:12.812 P-0028 T-002C HDCP SERVICE --- >0
[ 147.051190] TF : INFO 00000064 02:55:12.903 P-0028 T-002C HDCP SERVICE --- >1
[ 147.228315] nvavp nvavp: using SMMU at ff00000 to load AVP kernel
[ 147.234649] nvavp nvavp: AVP os at vaddr=f0e00000 paddr=ff00000 reset_addr=04
[ 147.311547] misc nvmap: nvmap_unpin_ids: mediaserver unpinning invalid handl0
[ 147.366602] request_suspend_state: wakeup (0->0) at 146775728003 (2013-06-07)
[ 147.377437] init: untracked pid 3903 exited
[ 147.414662] init: untracked pid 3902 exited

The last couple lines right before the loop, deal with hdmi stuff...

[ 10.490952] HDMI: detected monitor SCEI MONITOR at connection type HDMI
[ 10.497668] HDMI: available speakers: FL/FR
[ 10.509084] HDMI: supports coding type LPCM: channels = 2, rates = 44100 4804

Didn't really do any research yet with it, just ran it and posted... will look tomorrow.

 

[Vinny75]

just for the hell of it I grabbed the stable TF700t CM10.0 build from http://get.cm/?device=tf700t tried a quick port of that with the same results. I get a really cool CM boot logo to look at though!

Have you considered a clean port of CM to the OUYA instead of using the TF700t CM build? http://wiki.cyanogenmod.org/w/Doc:_porting_intro

I've been setting up my android environment and walking through this tutorial in prep for when I finally get my OUYA

edit: just saw post #42 on this...

 

[professorpoptart]

I unplugged the usb keyboard and micro usb, and it looks like that cleaned up the log a little. The looping part at the end of the log has to do with nvavp. Googling around it sounds like this has something to do with some kind of media service. I'm thinking maybe comment out whatever loads it to stop this from even loading and see if it continues... or get a different error...

Looking through the ramdisk of the merged boot for what loads this...

read firmware from 'nvavp_os_0ff00000.bin' (15516 b)
nvavp nvavp: read ucode firmware from 'nvavp_vid_ucode_alt.bin'

the files themselves are in /system/etc/firmware