The last few months I've spent so sporadically to create a layer for the BSP STLinux based set-top boxes. At the moment, the focus is very much on the Spark (Fulan) boxes, as I own a GM 990th There is already a definition for the Spark 7162 although this is untested. Other speakers, however, should not be too much effort.
Motivation
I wanted to because I have been concerned for some time with the theme OpenEmbedded Core / Yocto I thought I would look at times how much effort such a port would STLinux. I fully recognize that it is already the TDT, OpenWRT and soap's build system. They have for me each has its advantages and disadvantages. Until now I have from all the features like the build system of soap is best. It is clearly structured and offers a package management out of the box. For me Yocto offers a bit more than the system of soap, since it is somewhat better documented and a larger community behind it. Furthermore, it is once you understand the system easier to add packages. For me this is create a toolchain which also externally to develop third Party tools can be used simply priceless. I find layer concept now really super practical.
Current State
On my Golden Media 990 the core-image-minimal boots from the Yocto Project. The TDT drivers are loaded and the tools and how fp_control ustslave exist. The functionality of the driver is not tested yet, because I still have not translated GUI. It is indeed at the moment only a BSP layer. Yocto UDEV used as hotplug service therefore had a few udev rules for tools like ustslave be created. That and the use of the modutils should make it much easier Wi-Fi. Driver or other USB gadget driver from the desktop to the pit to port That I did in the case of Compat-Wireless driver for the Pinky TDT made almost insane.
Most of the time, the chain has cost me. In particular, the glibc because Yocto actually uses eglibc. For a time I wanted to work with an external toolchain, but that's kind of ugly. Then I dared them to install the glibc. That was easier than I thought it. I use the glibc git directly from the STLinux. When I use the GCC GCC which in denzil Branch of Yocto is used. In the case of the toolchain, it can feel good that is missing one or the other patch for the best possible performance. Me a bootable system was first important. If that works closed source PTI I have not tested too, because there is no GUI available.
Lirc works tested on my GM 990 at least with "irw". Including the activation of the front panel processor creating the "/ dev / vfd" seems to work. Whether all devices are properly built by UDEV I can only say after the first or any GUI tools are run as w_scan etc..
Next Steps
In addition to advancing the BSP layer I will start an applications layer. In this then the GUIs like neutrino, VDR and enigma are packed. With the focus at the moment on neutrino. Dawird still plenty of work waiting for me to compile all libraries and function.
Swell
Place all interested here at github under the name project magpie and the repository is called meta-STLinux
-
https://github.com/project-magpie/meta-stlinux -
-
https://github.com/project-magpie/meta-stlinux/wiki -
About patches Merge requests, suggestions and comments, I am always.
Thanks: 
Likes: 
Dislikes: 
Project Magpie
Reply With Quote
