aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--src/address/mod.rs3
-rw-r--r--src/lib.rs69
2 files changed, 62 insertions, 10 deletions
diff --git a/src/address/mod.rs b/src/address/mod.rs
index e44ca13..31f2079 100644
--- a/src/address/mod.rs
+++ b/src/address/mod.rs
@@ -68,6 +68,9 @@ impl<T: Address> AddressDiff<T> {
pub fn from_const(amount: T::Diff) -> Self {
AddressDiff { amount }
}
+ pub fn to_const(&self) -> T::Diff {
+ self.amount
+ }
}
impl<T: Address> fmt::Debug for AddressDiff<T> where T::Diff: fmt::Debug {
diff --git a/src/lib.rs b/src/lib.rs
index 1862993..121cd1b 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,19 +1,10 @@
#![no_std]
-#[cfg(feature = "std")]
-#[macro_use]
-extern crate std;
-
use core::fmt::{self, Debug, Display};
use core::hash::Hash;
-extern crate num_traits;
-#[cfg(feature="use-serde")]
-extern crate serde;
#[cfg(feature="use-serde")]
#[macro_use] extern crate serde_derive;
-#[cfg(feature="colors")]
-extern crate crossterm;
#[cfg(feature="use-serde")]
use serde::{Serialize, Deserialize};
@@ -32,14 +23,33 @@ pub use color::ColorSettings;
pub mod display;
mod reader;
-pub use reader::{Reader, ReadError, U8Reader}; //, U16le, U16be, U32le, U32be, U64le, U64be};
+pub use reader::{Reader, ReadError, U8Reader, U16le, U16be, U32le, U32be, U64le, U64be};
+/// the minimum set of errors a `yaxpeax-arch` disassembler may produce.
+///
+/// it is permissible for an implementor of `DecodeError` to have items that return `false` for
+/// all these functions; decoders are permitted to error in way that `yaxpeax-arch` does not know
+/// about.
pub trait DecodeError {
+ /// did the decoder fail because it reached the end of input?
fn data_exhausted(&self) -> bool;
+ /// did the decoder error because the instruction's opcode is invalid?
+ ///
+ /// this may not be a sensical question for some instruction sets - `bad_opcode` should
+ /// generally indicate an issue with the instruction itself. this is in contrast to one
+ /// specific operand being invalid for the instruction, or some other issue to do with decoding
+ /// data beyond the top-level instruction. the "opcode"/"operand" distinction is often fuzzy
+ /// and left as best-effort for decoder implementors.
fn bad_opcode(&self) -> bool;
+ /// did the decoder error because an operand of the instruction to decode is invalid?
+ ///
+ /// similar to [`DecodeError::bad_opcode`], this is a subjective distinction and best-effort on
+ /// the part of implementors.
fn bad_operand(&self) -> bool;
}
+/// a minimal enum implementing `DecodeError`. this is intended to be enough for a low effort,
+/// low-fidelity error taxonomy, without boilerplate of a `DecodeError` implementation.
#[derive(Debug, PartialEq, Eq, Copy, Clone)]
pub enum StandardDecodeError {
ExhaustedInput,
@@ -63,15 +73,43 @@ impl DecodeError for StandardDecodeError {
fn bad_operand(&self) -> bool { *self == StandardDecodeError::InvalidOperand }
}
+/// an interface to decode [`Arch::Instruction`] words from a reader of [`Arch::Word`]s. errors are
+/// the architecture-defined [`DecodeError`] implemention.
pub trait Decoder<A: Arch + ?Sized> {
+ /// decode one instruction for this architecture from the [`yaxpeax_arch::Reader`] of this
+ /// architecture's `Word`.
fn decode<T: Reader<A::Address, A::Word>>(&self, words: &mut T) -> Result<A::Instruction, A::DecodeError> {
let mut inst = A::Instruction::default();
self.decode_into(&mut inst, words).map(|_: ()| inst)
}
+ /// decode one instruction for this architecture from the [`yaxpeax_arch::Reader`] of this
+ /// architecture's `Word`, writing into the provided `inst`.
+ ///
+ /// SAFETY:
+ ///
+ /// while `inst` MUST be left in a state that does not violate Rust's safety guarantees,
+ /// implementors are NOT obligated to leave `inst` in a semantically meaningful state if
+ /// decoding fails. if `decode_into` returns an error, callers may find contradictory and
+ /// useless information in `inst`, as well as *stale data* from whatever was passed in.
fn decode_into<T: Reader<A::Address, A::Word>>(&self, inst: &mut A::Instruction, words: &mut T) -> Result<(), A::DecodeError>;
}
+/// a collection of associated type parameters that constitute the definitions for an instruction
+/// set. `Arch` provides an `Instruction` and its associated `Operand`s, which is guaranteed to be
+/// decodable by this `Arch::Decoder`. `Arch::Decoder` can always be constructed with a `Default`
+/// implementation, and decodes from a `Reader<Arch::Address, Arch::Word>`.
+///
+/// `Arch` is suitable as the foundational trait to implement more complex logic on top of; for
+/// example, it would be entirely expected to have a
+/// ```text
+/// pub fn emulate<A: Arch, E: Emulator<A>>(
+/// reader: &mut Reader<A::Address, A::Word>,
+/// emu: &mut E
+/// ) -> Result<A::Address, DecodeOrEvaluationError>;
+/// ```
+///
+/// in some library built on top of `yaxpeax-arch`.
#[cfg(feature="use-serde")]
pub trait Arch {
type Word: Debug + Display + PartialEq + Eq;
@@ -92,9 +130,20 @@ pub trait Arch {
type Operand;
}
+/// instructions have lengths, and minimum possible sizes for advancing a decoder on error.
+///
+/// unfortunately, this means calling `x.len()` for some `Arch::Instruction` requires importing
+/// this trait. sorry.
pub trait LengthedInstruction {
type Unit;
+ /// the length, in terms of `Unit`, of this instruction. because `Unit` will be a diff of an
+ /// architecture's `Address` type, this almost always is a number of bytes. implementations
+ /// should indicate if this is ever not the case.
fn len(&self) -> Self::Unit;
+ /// the length, in terms of `Unit`, of the shortest possible instruction in a given
+ /// architecture.. because `Unit` will be a diff of an architecture's `Address` type, this
+ /// almost always is a number of bytes. implementations should indicate if this is ever not the
+ /// case.
fn min_size() -> Self::Unit;
}