I will look into which is the last commit considered in the image and I will post it here
About the patches, one thing you could do is to look into the .patch and .do_patch files in the edison-src/build/tmp/work folder. Also take a look at the documentation for Yocto: Patching the Kernel - Yocto. I hope this works.
I think the instructions in the Patching the kernel are very similar to what I was thinking of doing, so I tried what I mentioned.
Took me two attempts as I forgot to redo the stuff for menu-config as to select the Sony HID stuff to be included. I think that part may now be actually working as now do some of the normal bluetooth stuff like:
rfkill unblock bluetooth
Plug the PS3 into USB and use the tool to bind the joystick.
use Bluetoothctl and do a scan and then do a trust and connect to the joystick, which appear to work.
However still now /dev/input/js0 object. Which I know is probably a different subject.. On that end, I have gone back to menu config and I think enable some of the joystick stuff including the generic analog joystick and did another build, another reflash of the board... (Wonder if there is way to simply find the new generated modules) and simply somehow install them without doing a flash... (Maybe again off this topic).
Still no /dev/input/js0 object. My next guess is I need to create some new udev rules. That is for example on my Ubuntu 14.04 machine the file /lib/udev/rules.d/60-persistent-input.rules.d has some rules to I believe create the js objects from both id links as well as by-path. These types of rules are not (yet) in the edison version of this file..
There are also some other rules for the js objects in a couple other rules files on my NUC. May try adding these same rules to see if this would help.
Again another joystick side note: wish I could also just do an opkg install of jstest (may ask in other thread). But I also have my own test programs.
Another Question I meant to ask, is there an set of steps that one can take to build an Edison Image directly using the github project? Maybe something like a setup script (or special parameter to make) and a make ?
Alternatively it would be great if Intel posted builds somewhere that matches what is up on github?
The last commit should be the 43befd916a21fcd96acf222e1da766012e085486
About dev/input/js0 how are you including the module? Are you installing it or are you adding it as a module? Which modules did you enabled?
How are you adding jstest right now? An alternative is to create the .ipk file of jstest but if you are building your own image it would be better to include it in the image too.
There isn't a step by step tutorial for how to build an image from the github project but I will pass your ideas to the proper team. Thank you for the feedback.
If you want, you can use the files from the github project to build the kernel you will also need the layers from https://git.yoctoproject.org/cgit/cgit.cgi/meta-intel-edison/log/
Take a look at the following document, it may help you with this: https://software.intel.com/sites/default/files/managed/f1/49/building_yocto_and__ipks.pdf
Was gone a few days, so will take a look. Thanks for the information showing which was the last thing included in the current release.
As for your question: I tried it both as <M> as well as <*> (Module and built-in)
As for which ones I included ( I would need to run it again to give you the exact ones. I also believe I may have mentioned them in the other thread about joysticks.
But it included things like (Joystick support, Generic Analog support and some Test one). I was not sure if that was jstest or not, but if not I have my own test program that is part of my Raspberry_pi project up on github (kurte).
Will have to take a look at the github project you showed in the last posting to see if I can setup to directly build from github, which would be great.
Do you have updates in this?
Sorry, I need to get back to this.
Warning: This is sort-of off my topic here...
I sort-of go around in cycles with the HR-OS1. Started off with RPI2, had issues with it talking reliably to the Arbotix Pro controller (arm M0 using FTDI to communicate). So went back to Edison, although not built in. There does not appear to be the same issues with FTDI. Not sure why... But was running into issues with PS3 controller... So I decided to try swapping in an Odroid C1+, which is the same size and has the same hole pattern of RPI2. Quad core 1.5ghz...
Run into same issue with talking to Arbotix Pro, but works fine with PS3 using sixad... So I started to diagnose it. It was timing out on waiting for responses. So far I found having the code that talks to the Arbotix Pro do a call to tcdrain on the file handle before it starts the timeout code waiting for a response has helped. I am no longer getting timeouts, but I do think there are still latency issues. (Latency set to 1 in FTDI driver)... Still diagnosing.
(Back on topic)
So soon I will try some more with the Edison. I have been sort-of procrastinating here, hoping that either a) someone comes out and says yep here are the steps, or potentially better yet, b) Intel releases a new build with all of the stuff in it, including SPI fixes, FTDI built in, ...
Thanks again Charlie,