Raspberry Pi images with Xenomai 3 on Raspbian linux

Images

Raspberry Pi image for the Pi 2,3b and 3b+ with Xenomai 3.0.8 on Raspbian linux 4.9.y Debian 9 stretch

image file for raspberry pi sdcard for the Pi 2,3b and 3b+:

   raspbian_stretch9_xenomai3.0.8.img.7z (created june 2019)

the instructions for building this image you can find at https://github.com/harcokuppens/xenomai3_rpi_gpio/blob/master/install/install_xenomai-3.0.8_on_rpi2_3_3B+.md.
I made an improved installation description for building the kernel with xenomai support at https://github.com/harcokuppens/xenomai3_rpi_gpio/blob/master/install/install_xenomai-3.0.8_on_rpi_kernel_v4.9.y_4.9.80__for_rpi_2till3bplus.md. This installation creates a single debian package
```
  raspberrypi-kernel-4.9.80-rpi-xeno-3.0.8_armhf.deb 
```
which you just have to install on a standard pi image to get xenomai support on that image. It installs the xenomai patched kernel, its modules and the xenomai tools and libraries in one go!
if you want to compile a kernel module using the linux headers you must also install the following package
```
  linux-headers-4.9.80-rpi-xeno-3.0.8+_4.9.80-rpi-xeno-3.0.8+-1_armhf.deb 
```
Note: if you want to compile a kernel module using these linux headers you must first run the following command:
```
   
              cd  /usr/src/linux-headers-4.9.80-rpi-xeno-3.0.8+/
              make -i modules_prepare
         
```
Ignore the errors, it just only needs to recompile the scripts in the linux headers folder. These scripts are in intel format in the package because I cross compiled these debian packages on an intel machine and somehow the scripts where compiled for the local platform instead of the arm platform, so therefore we need to recompile them on the pi it self.
I shrinked the image file with the very nice instructions from Andrew Oakley which you can find here.
This image is not perfect. Xenomai and handling gpio interrupts in realtime work fine on this image, however somehow handling gpio interrupts in linux don't work anymore. There is an issue with linux not getting gpio interrupts anymore in the patched kernel even when the xenomai module is not loaded. This problem is addressed on the xenomai mailinglist and seems to be an ipipe issue, but until now no fix is found. Anyhow if we just handle all gpio interrupts in xenomai we are fine with the image. But it still would be better if this issue was fixed.

Raspberry Pi image for the Pi zero,1,2,3b but not 3b+ with Xenomai 3.0.5 on Raspbian linux 4.1.y Debian 8 jessie

image file for raspberry pi sdcard for the Pi zero,1,2,3b but not 3b+:
```
   raspbian_jessie8_xenomai3.0.5.img.7z (created juli 2017) 
```
the instructions for building the kernel for the rpi 2 and rpi 3b for this image you can find at https://github.com/harcokuppens/xenomai3_rpi_gpio/blob/master/install/install_xenomai-3-3.0.5_on_rpi2__using_official_ipipe_patch__external_bcm_patch__pinctrl-bcm2835_patchfrom_ipipe_repos.txt.
the instructions for building the kernel for the rpi 1 for this image you can find at https://github.com/harcokuppens/xenomai3_rpi_gpio/blob/master/install/install_xenomai-3-3.0.5_on_rpi1__ipipe_git.txt.
I shrinked the image file with the very nice instructions from Andrew Oakley which you can find here.
I also compiled for this image a debian package to install the kernel and its headers on a standard pi image. For the pi1 these are linux-firmware-image-4.1.21-ipipe+_4.1.21-ipipe+-1_armhf.deb and linux-headers-4.1.21-ipipe+_4.1.21-ipipe+-1_armhf.deb. And for the pi2 and pi3 these are linux-firmware-image-4.1.21-v7+_4.1.21-v7+-4_armhf.deb and linux-headers-4.1.21-v7+_4.1.21-v7+-4_armhf.deb.
Note: if you want to compile a kernel module using the linux headers you must first run the following command:
```
    make -i modules_prepare
```
Ignore the errors, it just only needs to recompile the scripts in the linux headers folder. These scripts are in intel format in the package because I cross compiled these debian packages on an intel machine and somehow the scripts where compiled for the local platform instead of the arm platform, so therefore we need to recompile them on the pi it self.

Getting started with the image

Download the image, unzip it with 7zip and put the image on a sdcard. Then put the sdcard in the pi, and reboot. Voila : running pi with xenomai
Instructions on how to put the image on a sdcard you can find here.
you can log in with the following credentials:
```
      user:            pi
      passwd:          raspberry 

      user:            root
      passwd:          pi
      
```
For working with xenomai the root user is preferred.

Note: if as root you by accident destroy something, then you can just reinstall the original image and start fresh again.
When you use the new card for the first time, remember to expand the filesystem to fill the space on the new card. This because the image is made as small to just contain all it needs. However most sdcards are much bigger, but before you can use this extra space you must expand the filesystem first. From the Raspberry Pi's terminal, do sudo raspi-config , then select Expand Filesystem. Then do a reboot to apply the changes. Alternatively you can do this directly from the command line with:
```
    sudo raspi-config --expand-rootfs
    sudo shutdown -r now
```

Remote access

when to pi boots it does dhcp on both :
- ethernet : when in the lab the pi is connected with an utp cable to an utp socket it will fetch an address from the Radboud University WAN, so the pi becomes a WAN client in the network domain 131.174.*.*
- wifi : when used in the lab it will fetch an address on a local wifi hotspot which serves a local LAN, so the pi becomes a local LAN client in the network domain 168.192.*.*. The wifi router also does NAT, so the local clients can also access to the outside Radboud University WAN.
```
                     wifi settings: 

                         SSID: ASUS_NDL           2.4 gigahertz  PREFERRED
                         SSID: ASUS_5G_NDL        5 gigahertz

                         security: WPA2 personal
                         password: ask-for-it
                     
```
IMPORTANT:
- If you want to connect from your laptop/pc to a pi in the wifi LAN then also your laptop must be on the same wifi LAN, otherwise you cannot connect. The clients in the LAN are hidden from the WAN.
to find out the ip address of the pi you can go to the website below and look it up using either its wlan or ethernet mac address:

http://www.cs.ru.nl/lab/dyndns/
Note: at each boot the pi updates its ip address to the above server. The server also shows the time of update, so that you can see whether the entry is recent or old. If you connect an utp cable after booting you have to reboot the pi to have the ip send to the server. Also sometimes fetching an ip address is so slow for the first time that the dyndns update script is run before the new ip is fetched. On the next boot fetching the ip will be faster, so in that case just reboot. Luckily booting of the pi is very fast!!

IMPORTANT:
- The above server only allows clients from the Radboud University WAN. It also only stores ip-addresses from the Radboud University WAN ( 131.174.*.* ) and from the lab's wifi LAN (168.192.*.*) using NAT. In the latter case, using NAT causes the server see the LAN client with the ip of the wifi router, which is in the Radboud University WAN, and therefore the client is still allowed to update its own ip to the server.
- The dyndns server and client software is downloadable from : https://github.com/harcokuppens/dyndns
The idea is that if you know the mac addres of your pi, then you don't need to connect a monitor,mouse and keyboard to it, but instead you can lookup its ip address using above website, and login over ssh into the pi. Or smbmount the root homedir on your own machine so that you can easily edit the files on the pi with your local editor.

Note: just put a label with the mac address on your pi !!
to find out the mac and or ip address of a network interface on the pi, you can just connect a monitor a mouse and a keyboard to the pi, login to it , and run the following command :
```
/sbin/ifconfig
```
Note: at each login the pi automatically prints the mac and ip address of the eth0 and wlan0 interfaces
the next two services will be started :
1. ssh server : ssh root@<ip-address>
  With this server you can log in to a remote shell on the os running in VirtualBox. In this shell you can edit files with text editors like joe/nano/vim.
2. samba server :
  This server exports the root file system as networkshare for windows.
  On linux and macintosh computers you can access this share by going to the network location :
```
              smb://<ip-address>/root
              user: root
              passwd: pi
          
```
  On windows you can access this share by typing the following in the explorer's address bar :
```
              \\<ip-address>\root
              user: root        
              passwd: pi
          
```

Startup profile for xenomai

In the startup profile directory I added the script /etc/profile.d/xenomai.sh which at login:

prints the current xenomai version
fixes paths for xenomai (PATH and LD_LIBRARY_PATH)

adds several convenient aliases and functions to the bash environment :

               alias xeno-stat="cat /proc/xenomai/sched/stat"
               alias xeno-threads="cat /proc/xenomai/sched/threads"
               alias xeno-rt-threads="cat /proc/xenomai/sched/rt/threads"
               alias xeno-interrupts="cat /proc/xenomai/irq"
               alias xeno-version=/usr/xenomai/sbin/version

               # convenience methods to switch between cobalt and mercury
               # note: mercury is running on xenomai-cobalt patched kernel but without rt_preempt patch
               #        => running on mercury on this image is thus just running on linux but then as high prio (fifo) thread within linux (independent cobalt)
               #        => just to show portability xenomai programs from cobalt to mercury
               #        => important: most rtdm drivers are only available for cobalt and are not available yet for mercury  (eg. gpio rtdm driver)
               xeno-switch-cobalt() {
                 ...
               }
               xeno-switch-mercury() {
                 ...
               }

For the stretch image we don't have xenomai mercury but only xenomai cobalt installed, so for that image we don't have the xeno-switch commands.

Multiple consoles & Gui

When you have connected a monitor, keyboard and a mouse to the pi when the pi boots you get a terminal console.

you can use ctrl-alt-F1, ctrl-alt-F2, ..., ctrl-alt-F6 to switch between different consoles
you can also install the raspian Pixel Desktop :
```
    sudo apt-get update
    sudo apt-get dist-upgrade
    sudo apt-get install -y rpi-chromium-mods
    sudo apt-get install -y python-sense-emu python3-sense-emu python-sense-emu-doc
      
```
Just reboot the pi to login to do a graphical login to Pixel, which is on virtual console 7 (ctrl-alt-F7).
More info about the Pixel Desktop you can find at : introducing pixel