[][src]Crate bounded_registers

Bounded Registers

A high-assurance memory-mapped register code generation and interaction library.


$ git clone git@github.com:auxoncorp/registers.git
$ cd registers && cargo install


There are two core pieces to bounded-registers:

I. The macro

register! {
    Fields [
        On WIDTH(U1) OFFSET(U0),
        Dead WIDTH(U1) OFFSET(U1),
        Color WIDTH(U3) OFFSET(U2) [
            Red = U1,
            Blue = U2,
            Green = U3,
            Yellow = U4

The register! macro generates the code necessary for ergonomic register access and manipulation. The expected input for the macro is as follows:

  1. The register name.
  2. Its numeric type.
  3. Its mode, either RO (read only), RW (read write), or WO (write only).
  4. The register's fields, beginning with Fields [, and then a closing ] at the end.

A field constists of its name, its width, and its offset within the register. Optionally, one may also state enum-like key/value pairs for the values of the field, nested within the field declaration with []'s

The code which this macro generates is a tree of nested modules where the root is a module called $register_name. Within $register_name, there will be the register itself, as $register_name::Register, as well as a child module for each field.

Within each field module, one can find the field itself, as $register_name::$field_name::Field, as well as a few helpful aliases and constants.

  • $register_name::$field_name::Read: In order to read a field, an instance of that field must be given to have access to its mask and offset. Read can be used as an argument to get_field so one does not have to construct an arbitrary one when doing a read.
  • $register_name::$field_name::Clear: A field whose value is zero. Passing it to modify will clear that field in the register.
  • $register_name::$field_name::Set: A field whose value is $field_max. Passing it to modify will set that field to its max value in the register. This is useful particularly in the case of single-bit wide fields.
  • $register_name::$field_name::$enum_kvs: constants mapping the enum like field names to values.

II. Interacting with registers

Through a constructor

fn main() {
    let mut reg = Status::Register::new(0);
    assert_eq!(reg.read(), 2);

In this example, we initialize a register with the value 0 and then set the Dead bit—the second field—which should produce the value 2 when interpreting this word-sized register as a u32.

Through a register block

Here we take a known address, one we may find in a developer's manual, and interpret that address as a register block. We can then dereference that pointer and use the register API to access the registers in the block.

You can then implement Deref and DerefMut for a type which holds onto the address of such a register block. This fills in the gaps for method lookup (during typechecking) so that you can ergonomically use this type to interact with the register block:

extern crate bounded_registers;
extern crate typenum;

use core::ops::{Deref, DerefMut};

register! {
    Fields [
        Data        WIDTH(U8) OFFSET(U0),
        ParityError WIDTH(U1) OFFSET(U10),
        Brk         WIDTH(U1) OFFSET(U11),
        FrameError  WIDTH(U1) OFFSET(U12),
        Overrrun    WIDTH(U1) OFFSET(U13),
        Error       WIDTH(U1) OFFSET(U14),
        ChrRdy      WIDTH(U1) OFFSET(U15)

register! {
    Fields [
        Data WIDTH(U8) OFFSET(U0)

register! {
    Fields [
        Enable              WIDTH(U1) OFFSET(U0),
        Doze                WIDTH(U1) OFFSET(U1),
        AgingDMATimerEnable WIDTH(U1) OFFSET(U2),
        TxRdyDMAENable      WIDTH(U1) OFFSET(U3),
        SendBreak           WIDTH(U1) OFFSET(U4),
        RTSDeltaInterrupt   WIDTH(U1) OFFSET(U5),
        TxEmptyInterrupt    WIDTH(U1) OFFSET(U6),
        Infrared            WIDTH(U1) OFFSET(U7),
        RecvReadyDMA        WIDTH(U1) OFFSET(U8),
        RecvReadyInterrupt  WIDTH(U1) OFFSET(U9),
        IdleCondition       WIDTH(U2) OFFSET(U10),
        IdleInterrupt       WIDTH(U1) OFFSET(U12),
        TxReadyInterrupt    WIDTH(U1) OFFSET(U13),
        AutoBaud            WIDTH(U1) OFFSET(U14),
        AutoBaudInterrupt   WIDTH(U1) OFFSET(U15)

pub struct UartBlock {
    rx: UartRX::Register,
    _padding1: [u32; 15],
    tx: UartTX::Register,
    _padding2: [u32; 15],
    control1: UartControl1::Register,

pub struct Regs {
    addr: usize,

impl Deref for Regs {
    type Target = UartBlock;

    fn deref(&self) -> &UartBlock {
        unsafe { &*(self.addr as *const UartBlock) }

impl DerefMut for Regs {
    fn deref_mut(&mut self) -> &mut UartBlock {
        unsafe { &mut *(self.addr as *mut UartBlock) }

fn main() {
    let mut x = [0_u32; 33];
    let mut regs = Regs {
        // Some shenanigans to get at `x` as though it were a
        // pointer. Normally you'd be given some address like
        // `0xDEADBEEF` over which you'd instantiate a `Regs`.
        addr: &mut x as *mut [u32; 33] as usize,

    assert_eq!(regs.rx.read(), 0);
        .modify(UartControl1::Enable::Set + UartControl1::RecvReadyInterrupt::Set);

    // The first bit and the 10th bit should be set.
    assert_eq!(regs.control1.read(), 0b_10_0000_0001);

The Register API

The register API code is generated with docs, but you'll have to build the rustdoc documentation for your library that uses bounded-registers to be able to see it. For convenience, I've extrapolated it here:

/// `new` constructs a read-write register around the
/// given pointer.
fn new(init: Width) -> Self;

/// `get_field` takes a field and sets the value of that
/// field to its value in the register.
fn get_field<M: Unsigned, O: Unsigned, U: Unsigned>(
    f: Field<Width, M, O, U, Register>,
) -> Option<Field<Width, M, O, U, Register>>
    U: IsGreater<U0, Output = True> + ReifyTo<Width>,
    M: ReifyTo<Width>,
    O: ReifyTo<Width>,
    U0: ReifyTo<Width>;

/// `read` returns the current state of the register as a `Width`.
fn read(&self) -> Width;

/// `extract` pulls the state of a register out into a wrapped
/// read-only register.
fn extract(&self) -> ReadOnlyCopy<Width, Register>;

/// `is_set` takes a field and returns true if that field's value
/// is equal to its upper bound or not. This is of particular use
/// in single-bit fields.
fn is_set<M: Unsigned, O: Unsigned, U: Unsigned>(
    f: Field<Width, M, O, U, Register>,
) -> bool
    U: IsGreater<U0, Output = True>,
    U: ReifyTo<Width>,
    M: ReifyTo<Width>,
    O: ReifyTo<Width>;

// `Positioned` is a special trait that all fields implement, as
// well as a type used as an accumulator when reading from or
// writing to multiple fields. To use these functions with
// multiple fields, join them together with `+`. An `Add`
// implementation for fields has been provided for this purpose.

/// `matches_any` returns whether or not any of the given fields
/// match those fields values inside the register.
fn matches_any<V: Positioned<Width = Width>>(&self, val: V) -> bool;

/// `matches_all` returns whether or not all of the given fields
/// match those fields values inside the register.
fn matches_all<V: Positioned<Width = Width>>(&self, val: V) -> bool;

/// `modify` takes one or more fields, joined by `+`, and
/// sets those fields in the register, leaving the others
/// as they were.
fn modify<V: Positioned<Width = Width>>(&mut self, val: V);

/// `write` sets the value of the whole register to the
/// given `Width` value.
fn write(&mut self, val: Width);


bounded-registers employs values—specifically numbers—at the type-level in order to get compile time assertions on interactions with a register. Each field's width is used to determine a maximum value, and then reading from and writing to those fields is either checked at compile time, through the checked function, or is expected to carry a proof, which uses the aforementioned bound to construct a value at runtime which is known to not contravene it.





The register! macro generates the code necessary for ergonomic register access and manipulation. It is the crux of this crate. The expected input for the macro is as follows:



A field in a register parameterized by its mask, offset, and upper bound. To construct a field, its val must be ⩽ U::U32.


FieldDisj is short for Field Disjunction. It is a type which constitutes the intermediate result of the summing, or disjunct of two fields. It is not a type which one should use directly.




Positioned is a trait which is used to represent a value, be it a Field or simply a u32, as its value were it to be in position in its register.