New embedded assignment

[Mirabellensaft]

Hi Jonathan, can you check dk_template/src/lib.rs if the button and UARTE implementation are correct so I can write the exercise text?

Notes about the proposed exercise

BSC Exercise

In this exercise you will learn how to write a board support crate.
The bsc template will already contain the led and timer implementation.
The radio and USB/Power implementations will be deleted, because that just takes up unnecessary space and adds to confusion.

Learning goals

• implement buttons functionality
• uarte implementation
• impl blocks, associated functions, methods
• generate docs!

Steps

Write a button implementation

• add field in the board struct

• add struct for all buttons

• add struct for the single button

• Read docs, section 8.7 for info about pins and pin configuration

• add button bring up to board init

• add doc lines every where!

• add methods in impl block:

  • detect button push
  • debounce button function? like in knurling session, requires implementation of a second timer, just for this?

Write Uarte implementation

• add field to the board struct
• add struct for the instance, how to figure out what the type of the inner field is
• create instance in init, add baudrate, parity etc.
• add to instantiation of board struct
• impl fmt::Write for the Uarte struct, simple write does not work because of dma
• example code with button is not a good idea for the simple button implementation.

I think this is plenty for an hour.

[jonathanpallant]

This may be simpler using as_micros.

let mut micros = duration.as_micros();
while micros > u128::from(u32::MAX) {
    self.inner.delay_us(u32::MAX);
    micros -= u128::from(u32::MAX);
}
self.inner.delay_us(micros as u32);

[jonathanpallant]

This is a Cortex-M4, so I think fetch_add is available?

[jonathanpallant]

I don't like how this conjures up an RTC reference. If you don't own the RTC object, I would do a raw pointer read from a literal integer address.

[jonathanpallant]

Also, I don't like reading a static AtomicU32 through a *mut u32 pointer. It just seems unnecessarily unsafe.

[jonathanpallant]

I would just use a u64 or a u128 and let the compiler figure it out. It'd take me at least 10 minutes to satisfy myself this was correct and use most of a sheet of A4 paper doing it.

[jonathanpallant]

This "read it twice, and if before == after, then you're OK" is OK and a common idiom. That can stay.

[jonathanpallant]

Worth a note here as to which chip we're running on, and a link to the Manufacturer's data sheet / product page, so you can easily verify that these numbers are correct.

[jonathanpallant]

Add rerun-if-changed on memory.x?

[jonathanpallant]

I don't think you need these any more?

See https://defmt.ferrous-systems.com/migration-02-03.html