Why OS
======

.. toctree::
   :maxdepth: 1

   why-os/index

Course Structure
----------------

.. toctree::
   
   overview/index
   interrupt/index
   spinlocks/index
   threads/index

Todo
----

* Restructure why/code -> why/why-os

  * rename cmake targets accordingly

* Reformulate
  :doc:`/trainings/material/soup/linux/sysprog/scheduling/processes-and-threads`
  in terms of this

* Enable/disable safeguards against blocking in interrupts

  * ``class FD`` to handle this
  * start unchecked

    * inside the isr, prompt the user 
    * see how everything stalls
    * user input -> continue

  * enable check

Random Notes
------------

jjj sort that out

Intro
.....

When you somewhere want to consume data that was produced by an ISR,
you either do it in another (timer possibly) interrupt, or in the main
loop, pointlessly polling for something that will only happen once in
a lifetime. must do powermanagement in that loop, suspending the
device in deep sleep when on battery.

* show isr race condition on a list for example jjj link to source
* show fix via spinlock (show why-spinlock.cpp)

* Discussion: do much in ISR -> write program that has a global
  ``std::list``, and adds/removes one node in each isr

  https://codeberg.org/jfasch/jf-why-rtos/src/branch/main/firmware/do-much-in-isr.cpp

* Some :doc:`memory model </trainings/material/soup/cxx/cxx11/multithreading/memory-model/topic>`
* Add spinlock

  https://codeberg.org/jfasch/jf-why-rtos/src/branch/main/firmware/do-much-in-isr-spinlock.cpp

* Add why-shell

Scheduling, Tasks/Threads
.........................

Realtime
........

* hmmm show priority inversion?

Interrupt Communication
.......................

* message queue/pipe usage -> blocking in isr forbidden
* possibly only signal someone who's sleeping?
* one msqg for (thread, interrupt)? or ``msgq1=(thread1, (interrupt1, interrupt2)), msgq2=(thread2, (interrupt3))``

Polling
.......