XBMC on Tegra2 development: Porting XBMC for Linux to NVIDIA Tegra 2 ARM SoC platform

  Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Post Reply
akr884 Offline
Junior Member
Posts: 33
Joined: Apr 2010
Reputation: 0
Post: #31
davilla Wrote:$$ Just does not work out that way. From my experience with CrystalHD and adding a "donate" link on the google code page, user donations are dismal compared to the number of people running CrystalHD and if I had to depend on those to live, I'd starve.

Wouldn't it be better if it were on the xbmc.org donate page? I mean a system that announced it as a project priority, so that it wouldn't be limited to people who dig around in forums and follow the link trail away from the xbmc.org page. It sucks that people don't support you guys more (and I'm a total hypocrite, but I'm an unemployed college grad), but I mean trying to raise money for development goals without burying the development goals in forums and google code pages that can seem miles/km away from the main project front page.
find quote
theuni Offline
Team-XBMC Communication Manager
Posts: 1,105
Joined: Oct 2007
Reputation: 2
Location: Atlanta, Ga, USA
Post: #32
This has been discussed ad infinitum and we have decided on each occasion not do it. In theory it's a nice idea, but it's one that doesn't pan out in practice (with our program/community anyway).

If you really wish to discuss this, please do it in appropriate section of the forum. This is meant for dev only.

TheUni
find quote
akr884 Offline
Junior Member
Posts: 33
Joined: Apr 2010
Reputation: 0
Post: #33
Sorry to butt in here. I knew this wasn't the place for it, but I'm a little desperate for feedback before I try to develop this idea.

Would you mind if I copied what you said over to the thread I have on the general topics forum before you moderate my comments into oblivion?

And just for the record, I think it would be an idea I would want to talk to sourceforge or some meta-project about so that individual projects could put in place with minimal effort - so that each project wouldn't have to re-invent the wheel and maintain their system. Just trying to get feedback here because I use XBMC a lot and really appreciate the work that you all do.
find quote
mzsy Offline
Junior Member
Posts: 8
Joined: Apr 2010
Reputation: 0
Post: #34
davilla Wrote:sudo apt-get install build-essential

export CC=/usr/bin/gcc-4.5
export CXX=/usr/bin/g++-4.5

Thanks a lot ,it works..Something is different from sbox
find quote
jordan83 Offline
Junior Member
Posts: 1
Joined: Jun 2010
Reputation: 0
Post: #35
Hi all,

I'm trying to run XBMC on Tegra2 boards, unfortunately with little success so far :|

Here is what I've done...

1) I have installed Ubuntu karmic (9.10) on the board following this guide http://www.elinux.org/Installation_Detail and everything went pretty much ok.

2) I've compiled XBMC on the board, I followed the instructions that I found in README.armel in the svn repository. I concluded this part withouth any particular problems.
The configure string I used is:
Quote:./configure --disable-pulse --enable-external-python --enable-gles --enable-tegra

When I attempt to execute XBMC it complains that bpp is 16 instead of the minimum value of 24. How can I solve this issue? I tried to force a value of 24 bpp but it didn't work ^^

I'm interested in helping out (coding, testing your releases) with porting XBMC on tegra 2 but I need a bit of guidance to start off.

Nevertheless XBMC truly rocks, keep up the good work Smile

Have a nice day!!
(This post was last modified: 2010-06-25 12:39 by jordan83.)
find quote
McGeagh Offline
Team-XBMC ARM Developer
Posts: 178
Joined: Oct 2008
Reputation: 0
Location: Cambridge, UK
Post: #36
atleast for beagleboard, to which i had to run in 16bpp mode, all i did was run xbmc.bin directly, and not use the xbmc script in /usr/bin. could try that? (Alternatively you can edit the script (FEH) to allow 16bpp mode to be run)

¤ [McGeagh] ¤
find quote
vane.xbmc Offline
Member
Posts: 70
Joined: Feb 2010
Reputation: 0
Location: Spain
Post: #37
Hi all,

I'm trying to compile XBMC on Tegra 2 platform and it doesn't compile because of the following error:

OverlayRendererGL.cpp: In constructor OVERLAY::COverlayTextureGL::COverlayTextureGL (CDVDOverlayImage*): 128: error: GL_BGRA was not declared in this scope.

I'm configuring XBMC Dharma 10.0 with:
./configure --enable-external-python --enable-gles --disable-optimizations --enable-tegra --enable-openmax

Do you get this issue anytime?

Thanks in advance.

Regards,
Vanesa.
find quote
McGeagh Offline
Team-XBMC ARM Developer
Posts: 178
Joined: Oct 2008
Reputation: 0
Location: Cambridge, UK
Post: #38
It is erroring because OpenGL ES 2.0 does not support BGRA colourspace, only RGBA.

Either quickly change it to that (the shader should do bgra->rgba conversion anyways) and it should be happy... or wait for someone with time to fix (or do it yourself and post a patch!)

¤ [McGeagh] ¤
find quote
vane.xbmc Offline
Member
Posts: 70
Joined: Feb 2010
Reputation: 0
Location: Spain
Post: #39
Thanks McGeah for your quick answer.

I thought that XBMC Dharma was ready to compile for OpenGL ES, I don't understand the reason for this error. Am I wrong?

Regards,
Vanesa.
find quote
McGeagh Offline
Team-XBMC ARM Developer
Posts: 178
Joined: Oct 2008
Reputation: 0
Location: Cambridge, UK
Post: #40
well, it should be yes...
its possible that since last ES test, something got added or changed, which broke it for ES.
Ill put it on my todo list, but that list is very extensive so dont hold your breath.

¤ [McGeagh] ¤
find quote
vane.xbmc Offline
Member
Posts: 70
Joined: Feb 2010
Reputation: 0
Location: Spain
Post: #41
McGeagh Wrote:well, it should be yes...
its possible that since last ES test, something got added or changed, which broke it for ES.
Ill put it on my todo list, but that list is very extensive so dont hold your breath.

I understand.

I have just changed BGRA to RGBA and it compiles but I get an issue with X server when I try to run XBMC. It is described here: http://developer.nvidia.com/tegra/forum/...mment-5531.

Also, I ran xbmc.bin instead of xbmc as you said before in this thread because of the complaint about 24 bits.

Regards,
Vanesa.
find quote
McGeagh Offline
Team-XBMC ARM Developer
Posts: 178
Joined: Oct 2008
Reputation: 0
Location: Cambridge, UK
Post: #42
Without spending too much time on this, try doing the following:

via ssh, do the following:
export DISPLAY=:0
xinit -- -layout=HDMI &
(replacing HDMI with CRT if your hooked up via VGA)

This should boot x (takes a while) and then u can bring up a terminal in that GUI and launch ./xbmc.bin

Does that work?

¤ [McGeagh] ¤
find quote
vane.xbmc Offline
Member
Posts: 70
Joined: Feb 2010
Reputation: 0
Location: Spain
Post: #43
McGeagh Wrote:export DISPLAY=:0
xinit -- -layout=HDMI &
Does that work?

I have a serial connection with tegra2 and I have also started the X server with xinit. If I do this, I get a xterm in the HDMI display but I can not use the mouse or keyboard so I use the serial port to run xbmc.

If I run XBMC starting the X server with xinit I get this: http://pastebin.com/HDL0JjPE (I have to stop it with Ctrl + C).

Regards,
Vanesa.
find quote
McGeagh Offline
Team-XBMC ARM Developer
Posts: 178
Joined: Oct 2008
Reputation: 0
Location: Cambridge, UK
Post: #44
hmm, invalid gui shader doesnt look good, whats your xbmc.log say?

id love to look into this, but i really dont have the time atm. (by that, i mean im working on something else with my TG2 so cant test this as its a ballache to set it up for xbmc to test, then back again for what im currently working on)

¤ [McGeagh] ¤
find quote
vane.xbmc Offline
Member
Posts: 70
Joined: Feb 2010
Reputation: 0
Location: Spain
Post: #45
McGeagh Wrote:whats your xbmc.log say?

Here is the log: http://pastebin.com/2s7eqhtZ

GL_VENDOR and others are NULL. I remember this error on beagleboard. The problem was the flag SDL_OPENGL in WinSystemEGL.cpp, but this issue is solved en Dharma in this way:

#if (HAS_GLES == 2)
int options = 0;
#else
int options = SDL_OPENGL;
#endif

Regards,
Vanesa.
find quote