I don't have a /dev/fb0 on the beaglebone. I tried using the pre-built root file system in the TI bb SDK, and also rebuilding the TI SDK without any modification. Is there some trick here?
A bare BeagleBone does not have a display connected. So there is no /dev/fb0.
But the SDK still builds and launches the matrix app, right? And the beaglebone does have connectivity for video output (such as might be used for a DVI cape), right? Or what about connection through something like VNC?
The kernel detects which TI hardware platform it is on, EVM, beaglebone. Since there is not native display hardware on the beaglebone fb0 is not set up. Matrix is running though and can be accessed through a browser using the ip address of the beaglebone. However, please note that graphic related demos will not run in the browser.
The user will have to get the DVI sources from the cape provider and build them into the kernel to enable DVI.
The framebuffer would be registered once a DVI cape is detected (right now the SDK does not support any cape hardware). Registering a framebuffer device without the right cape being detected would not be correct since a different cape could use the same pins for a non-display purpose. In such a case you will end-up with a non-functional device node.
I have to admit I have not tried VNC to a beaglebone, but it seems VNC to a system without graphical display should be possible. I found some links like this which describe the process: http://stevenharman.net/blog/archive/2008/12/13/vnc-to-a-headless-ubuntu-box.aspx. Did you face any errors when trying VNC to BeagleBone?
I actually haven't tried the VNC either - I was just trying to think of scenarios where one would want to compose a picture using a beaglebone. I have a DVI cape, and have been using buildroot to build an RFS with gstreamer, gst-ffmpeg, etc., with the Angstrom u-boot and uImage. This lets me see H.264-doceded video on a DVI output, but the gstreamer ffmpegcolorspace (for YUV-to-RGB) and transfer to fbdevsink takes almost as long as the NEON-enabled H.264 decoding, so I wanted to look into using the SGX to do some of that. That led me to compiling the kernel with the TI SDK so that I could get the EGL stuff, which led me to discovering that there wasn't even a frame buffer when I used the TI SDK.
Thanks for the clarification. As Schuyler suggested, the best way forward would be to port the DVI cape patches from the angstorm kernel to the SDK kernel.
Bringing over the angstrom beagleboard did in fact allow my DVI cape to be recognized, create a /dev/fb0, and view the matrix app on a display device. The Qt demos appear to be working, and the video decode demos. The OpenGL demos didn't work, until I manually copied the .ko files from the linux-devkit. I guess the Graphics_SDK_setuplinux_4_06_00_01.bin installation is a separate step from the SDK compile.
Thanks for the help!
All content and materials on this site are provided "as is". TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with regard to these materials, including but not limited to all implied warranties and conditions of merchantability, fitness for a particular purpose, title and non-infringement of any third party intellectual property right. TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with respect to these materials. No license, either express or implied, by estoppel or otherwise, is granted by TI. Use of the information on this site may require a license from a third party, or a license from TI.
TI is a global semiconductor design and manufacturing company. Innovate with 100,000+ analog ICs andembedded processors, along with software, tools and the industry’s largest sales/support staff.