Merge pull request #52 from ianichitei/master

Add Rust bindings
pull/53/head
vlutas 3 years ago committed by GitHub
commit 62cdbdc068
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -63,7 +63,7 @@ jobs:
python3 -m pip install setuptools
- name: Build pybddisasm
run: |
cd pybddisasm
cd bindings/pybddisasm
python3 setup.py build
cd -
- name: Create package

@ -0,0 +1,39 @@
name: Rust
on:
push:
branches: [ master ]
pull_request:
branches: [ master ]
env:
CARGO_TERM_COLOR: always
jobs:
build:
strategy:
fail-fast: false
matrix:
os: [ windows-latest, macos-latest, ubuntu-latest ]
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v2
- name: Set LIBCLANG_PATH
run: echo "LIBCLANG_PATH=$((gcm clang).source -replace "clang.exe")" >> $env:GITHUB_ENV
if: matrix.os == 'windows-latest'
- name: Debug Build
run: cargo build --verbose --manifest-path bindings/rsbddisasm/Cargo.toml
- name: Release build
run: cargo build --release --verbose --manifest-path bindings/rsbddisasm/Cargo.toml
- name: Run tests
run: cargo test --verbose --manifest-path bindings/rsbddisasm/Cargo.toml
- name: Run clippy
run: cargo clippy --manifest-path bindings/rsbddisasm/Cargo.toml
- name: Run format checks
run: |
cd bindings/rsbddisasm
cargo fmt -- --check
cd ..

8
.gitignore vendored

@ -47,9 +47,6 @@ _UpgradeReport_Files/*
*.pyc
.*.swp
*.pyd
pydis/pydis.egg-info/
pydis/build/
pydis/dist/
.vs/
disasmtool_lix/build
@ -64,6 +61,9 @@ bdshemu_fuzz/out-32
bdshemu_fuzz/out-64
docs/build
libbddisasm.pc
build*
build/
.vscode
disasmtool_lix/_build
bindings/rsbddisasm/target
bindings/rsbddisasm/Cargo.lock
bindings/pybddisasm/pybddisasm.egg-info

@ -4,18 +4,13 @@ The Bitdefender disassembler (bddisasm) is a lightweight, x86/x64 only instructi
## Projects
1. bddisasm - this is the main disassembler project. In order to use the Bitdefender disassembler, all you have to do is build this project, and link with the output library. The only headers you need are located inside the `inc` folder.
2. bdshemu - this project makes use of the main bddisasm lib in order to build a simple, lightweight, fast, instructions emulator, designated to target shellcodes. This project is also integrated inside the disasmtool, so you can
1. [bddisasm](https://github.com/bitdefender/bddisasm/tree/master/bddisasm) - this is the main disassembler project. In order to use the Bitdefender disassembler, all you have to do is build this project, and link with the output library. The only headers you need are located inside the `inc` folder.
2. [bdshemu](https://github.com/bitdefender/bddisasm/tree/master/bdshemu) - this project makes use of the main bddisasm lib in order to build a simple, lightweight, fast, instructions emulator, designated to target shellcodes. This project is also integrated inside the disasmtool, so you can
emulate raw binary files, and see their output. Note that this simple emulator supports basic x86/x64 instructions, and does not support emulating any kind of API call. In addition, the only supported memory accesses are inside the shellcode itself, and on the emulated stack.
3. isagenerator - this project contains the instruction definitions and the scripts required to generate the disassembly tables. If you wish to add support for a new instruction, this is the place. This project will automatically generate several header files (instructions.h, mnemonics.h, constants.h, table_\*.h), so please make sure you don't manually edit any of these files. You will need Python 3 to run the generation scripts.
4. disasmtool - this project is a command line disassembler tool, used mainly as an example of how to integrate the bddisasm and bdshemu libraries.
5. disasmtool_lix - like disasmtool, but for Linux.
6. pybddisasm - this is the Python binding for the bddisasm project. You will need Python 3 for this.
3. [isagenerator](https://github.com/bitdefender/bddisasm/tree/master/isagenerator) - this project contains the instruction definitions and the scripts required to generate the disassembly tables. If you wish to add support for a new instruction, this is the place. This project will automatically generate several header files (instructions.h, mnemonics.h, constants.h, table_\*.h), so please make sure you don't manually edit any of these files. You will need Python 3 to run the generation scripts.
4. [disasmtool](https://github.com/bitdefender/bddisasm/tree/master/disasmtool) - this project is a command line disassembler tool, used mainly as an example of how to integrate the bddisasm and bdshemu libraries.
5. [disasmtool_lix](https://github.com/bitdefender/bddisasm/tree/master/disasmtool_lix) - like disasmtool, but for Linux.
6. [bindings](https://github.com/bitdefender/bddisasm/tree/master/bindings) - bindings for [python](https://github.com/bitdefender/bddisasm/tree/master/bindings/pybddisasm), and [Rust](https://github.com/bitdefender/bddisasm/tree/master/bindings/rsbddisasm).
## Objectives

@ -96,6 +96,7 @@ static char *TupleTypeToString(ND_TUPLE TupleType)
case ND_TUPLE_None: return "None";
case ND_TUPLE_FV: return "FV";
case ND_TUPLE_HV: return "HV";
case ND_TUPLE_QV: return "QV";
case ND_TUPLE_T1S8: return "T1S8";
case ND_TUPLE_T1S16: return "T1S16";
case ND_TUPLE_T1S: return "T1S";

@ -20,7 +20,7 @@ requires = ['setuptools']
here = os.path.abspath(os.path.dirname(__file__))
def _check_library_version():
version_header = '../inc/version.h'
version_header = '../../inc/version.h'
with open(version_header, 'r') as file:
data = file.read()
@ -92,8 +92,8 @@ setup(
extra_compile_args = ["-march=westmere"],
sources = ["_pybddisasm/_pybddisasm.c", "_pybddisasm/pybddisasm.c"],
define_macros = [('AMD64', None), ('LIBRARY_INSTRUX_SIZE', LIBRARY_INSTRUX_SIZE)],
include_dirs = ['../inc'],
include_dirs = ['../../inc'],
libraries = ['bddisasm'],
library_dirs = ['/usr/local/lib', '../bin/x64/Release'])],
library_dirs = ['/usr/local/lib', '../../build', '../../bin/x64/Release'])],
distclass=BinaryDistribution
)

@ -0,0 +1,6 @@
[workspace]
members = [
"bddisasm-sys",
"bddisasm",
]

@ -0,0 +1,28 @@
# bddisasm Rust bindings
## Build
Run `cargo build` or `cargo build --release`.
## Directory structure
### bddisasm-sys
This crate uses the [*-sys crate convention](https://doc.rust-lang.org/cargo/reference/build-scripts.html#-sys-packages) and links with `libbddisasm`, exposing a raw, low-level, interface.
### bddisasm
This crate aims to offer a higher-level interface over [bddisasm-sys](#bddisasm-sys).
Parts of it are auto-generated, with slight manual changes (for example, the `Mnemonic` enum and related functions).
#### TODO
- [ ] encode registers, not just the register index (for example, instead of `Gpr` we should have `Gpr::Rax`, `Gpr::Eax`, etc)
- [ ] more documentation for the `operand` module, with examples on how to work wit h each operand type
- [ ] more examples for `cpuid`
- [ ] more examples for `cpu_modes`
- [ ] an API to check if an instruction is supported with certain CPU modes on or off (currently a user has to manually check the `CpuModes` structure)
- [ ] implement `Display` for more types (especially those in `operand`)
- [ ] remove the `bindgen` dev dependency? See [this issue](https://github.com/rust-lang/rust-bindgen/issues/918) as to why we may want to do that
- [ ] better directory structure

@ -0,0 +1,26 @@
[package]
name = "bddisasm-sys"
version = "0.1.0"
authors = ["Cristi Anichitei <ianichitei@bitdefender.com>"]
edition = "2018"
links = "bddisasm"
build = "build.rs"
license = "Apache-2.0"
readme = "README.md"
repository = "https://github.com/bitdefender/bddisasm"
documentation = "https://docs.rs/bddisasm-sys"
description = """
Bindings to bddisasm instruction decoder library
"""
categories = ["external-ffi-bindings", "hardware-support"]
keywords = ["disassembler", "decoder", "x86", "amd64", "x86_64", "bindings"]
[lib]
name = "bddisasm_sys"
path = "src/lib.rs"
[dependencies]
[build-dependencies]
bindgen = "0.59.1"
cc = "1.0.70"

@ -0,0 +1,9 @@
# bddisasm-sys
Rust bindings for [bddisasm](https://github.com/bitdefender/bddisasm).
See [bddisasm](https://crates.io/crates/bddisasm) if you're looking for a Rust wrapper for these bindings.
## Requirements
[bindgen](https://crates.io/crates/bindgen) is used to generate the bindings at build time. Because of this, users need to have `clang` installed. Check the [bindgen documentation](https://rust-lang.github.io/rust-bindgen/requirements.html) for more information.

@ -0,0 +1,39 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
use std::env;
use std::path::PathBuf;
fn main() {
println!("cargo:rerun-if-changed=csrc");
cc::Build::new()
.file("csrc/bddisasm/bddisasm.c")
.file("csrc/bddisasm/bdformat.c")
.file("csrc/bddisasm/crt.c")
.include("csrc/bddisasm/include")
.include("csrc/inc")
.define("BDDISASM_HAS_VSNPRINTF", Some("1"))
.define("BDDISASM_HAS_MEMSET", Some("1"))
.compile("bddisasm");
let bindings = bindgen::Builder::default()
.header("csrc/bddisasm_wrapper.h")
.allowlist_function("Nd.*")
.allowlist_type("ND.*")
.allowlist_var("ND.*")
.rustified_enum(".*")
.impl_debug(true)
.generate_comments(false)
.parse_callbacks(Box::new(bindgen::CargoCallbacks))
.generate()
.expect("Unable to generate bindings");
let out_path = PathBuf::from(env::var("OUT_DIR").unwrap());
bindings
.write_to_file(out_path.join("bindings.rs"))
.expect("Couldn't write bindings!");
println!("cargo:rustc-link-lib=static=bddisasm");
}

@ -0,0 +1,10 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
#include <stdbool.h>
#include <stdint.h>
#include <stdlib.h>
#include "inc/bddisasm.h"

@ -0,0 +1,102 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Rust bindings for [bddisasm](https://github.com/bitdefender/bddisasm).
//!
//! See [bddisasm](https://crates.io/crates/bddisasm) if you're looking for a Rust wrapper for these bindings.
//!
//! # Requirements
//!
//! [bindgen](https://crates.io/crates/bindgen) is used to generate the bindings at build time. Because of this, users
//! need to have `clang` installed.
//! Check the [bindgen documentation](https://rust-lang.github.io/rust-bindgen/requirements.html) for more information.
#![allow(non_camel_case_types)]
#![allow(non_snake_case)]
#![allow(non_upper_case_globals)]
#![allow(deref_nullptr)] // See https://github.com/rust-lang/rust-bindgen/issues/1651
#![allow(clippy::all)]
include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
#[cfg(test)]
mod tests {
use super::*;
use ::std::os::raw::c_char;
use std::mem;
#[test]
fn get_version() {
let mut major: u32 = 0;
let mut minor: u32 = 0;
let mut revision: u32 = 0;
let mut build_date: *mut c_char = std::ptr::null_mut();
let mut build_time: *mut c_char = std::ptr::null_mut();
unsafe {
NdGetVersion(
&mut major,
&mut minor,
&mut revision,
&mut build_date,
&mut build_time,
);
}
println!("major: {} minor: {} rev: {}", major, minor, revision);
let build_date = unsafe { std::ffi::CStr::from_ptr(build_date) };
let build_date = build_date.to_str().unwrap().to_string();
let build_time = unsafe { std::ffi::CStr::from_ptr(build_time) };
let build_time = build_time.to_str().unwrap().to_string();
println!("Build date: {} build time: {}", build_date, build_time);
// There are no other asserts in this test. Enforcing a known minor version is not worth it, we mainly want to
// see that `NdGetVersion` works.
assert_eq!(major, 1);
}
fn do_decode(code: &[u8]) -> (INSTRUX, NDSTATUS) {
let mut instrux: mem::MaybeUninit<INSTRUX> = mem::MaybeUninit::uninit();
let instrux = instrux.as_mut_ptr();
let status = unsafe {
NdDecodeEx(
instrux,
code.as_ptr(),
code.len() as u64,
ND_CODE_32 as u8,
ND_DATA_32 as u8,
)
};
(unsafe { *instrux }, status)
}
#[test]
fn decode() {
let code = vec![0x90];
let (instrux, status) = do_decode(&code);
assert_eq!(status, 0, "Failed to decode instruction {:#x?}", code);
assert_eq!(
unsafe { instrux.__bindgen_anon_2.Instruction },
_ND_INS_CLASS::ND_INS_NOP
);
}
#[test]
fn format() {
let code = vec![0x89, 0x29];
let (instrux, status) = do_decode(&code);
assert_eq!(status, 0, "Failed to decode instruction {:#x?}", code);
let mut buffer: [i8; ND_MIN_BUF_SIZE as usize] = [0; ND_MIN_BUF_SIZE as usize];
let status = unsafe { NdToText(&instrux, 0, ND_MIN_BUF_SIZE, buffer.as_mut_ptr()) };
assert_eq!(status, 0, "Failed to decode format {:#x?}", code);
let text = String::from_utf8(buffer.iter().map(|&c| c as u8).collect()).unwrap();
let text = text.trim_matches(char::from(0));
assert_eq!(text, "MOV dword ptr [ecx], ebp");
}
}

@ -0,0 +1,17 @@
[package]
name = "bddisasm"
version = "0.1.0"
authors = ["Cristi Anichitei <ianichitei@bitdefender.com>"]
edition = "2018"
license = "Apache-2.0"
readme = "README.md"
repository = "https://github.com/bitdefender/bddisasm"
documentation = "https://docs.rs/bddisasm"
description = """
Bindings to bddisasm instruction decoder library
"""
categories = ["api-bindings", "hardware-support"]
keywords = ["disassembler", "decoder", "x86", "amd64", "x86_64", "bindings"]
[dependencies]
bddisasm-sys = { version = "0.1.0", path = "../bddisasm-sys" }

@ -0,0 +1,136 @@
# bddisasm
Rust bindings for the [bddisasm](https://github.com/bitdefender/bddisasm) x86/x64 decoder library, built on top
of [bddisasm-sys](https://crates.io/crates/bddisasm-sys).
It supports all existing x86 instruction, offering a wide range of information about each one, including:
- operands (implicit and explicit)
- access mode for each operand
- CPUID feature flags
- CPU modes in which an instruction is valid
## Usage
Add `bddisasm` to your `Cargo.toml`:
```toml
[dependencies]
bddisasm = "0.1.0"
```
## Examples
### Decoding one instruction
```rust
use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode, Mnemonic};
let code = vec![0x31, 0xc0];
match DecodedInstruction::decode(&code, DecodeMode::Bits32) {
Ok(ins) => {
assert_eq!(ins.mnemonic(), Mnemonic::Xor);
println!("{}", ins);
},
Err(err) => println!("Unable to decode: {}", err),
}
```
### Decoding multiple instructions
```rust
use bddisasm::decoder::{Decoder, DecodeMode};
let code = [
// ENCLS
0x0f, 0x01, 0xcf,
// MOV rax, qword ptr [rbx+rcx*4+0x1234]
0x48, 0x8b, 0x84, 0x8b, 0x34, 0x12, 0x00, 0x00,
// Not a valid instruction
0x0f,
// WRMSR
0x0f, 0x30,
];
let decoder = Decoder::new(&code, DecodeMode::Bits64, 0x1234);
for ins in decoder {
match ins {
Ok(ins) => println!("{}", ins),
Err(e) => println!("{}", e),
}
}
```
This will print:
```text
ENCLS
MOV rax, qword ptr [rbx+rcx*4+0x1234]
the provided input buffer is too small
WRMSR
```
### Working with instruction operands
Rich informaion is offered for each type of operand. Bellow is a minimal example that looks at a memory operand.
```rust
# use bddisasm::decode_error::DecodeError;
# fn test() -> Result<(), DecodeError> {
use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode};
use bddisasm::operand::OpInfo;
// ` MOV rax, qword ptr [rcx+r15*2]`
let code = b"\x4a\x8b\x04\x79";
let ins = DecodedInstruction::decode(code, DecodeMode::Bits64).unwrap();
// Get the operands
let operands = ins.operands();
// Get the second operand which is the source (`[rcx+r15*2]`)
let src = operands[1];
println!("Source operand type: {}", src.info);
match src.info {
OpInfo::Mem(mem) => {
if let Some(base) = mem.base {
println!("Base register: {}", base);
} else {
println!("No base register");
}
if let Some(index) = mem.index {
println!("Index register: {}", index);
} else {
println!("No index register");
}
if let Some(scale) = mem.scale {
println!("Scale: {}", scale);
} else {
println!("No scale");
}
if let Some(displacement) = mem.disp {
println!("Displacement: {}", displacement);
} else {
println!("No displacement");
}
},
_ => unreachable!(),
}
# Ok(())
# }
```
Will print:
```text
Source operand type: memory
Base register: 1
Index register: 15
Scale: 2
No displacement
```
## Requirements
Because [bddisasm-sys](https://crates.io/crates/bddisasm-sys) uses [bindgen](https://crates.io/crates/bindgen) to generate the bindings at build time, users need to have `clang` installed. Check the [bindgen documentation](https://rust-lang.github.io/rust-bindgen/requirements.html) for more information.

@ -0,0 +1,147 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Offers information about the CPU modes in which an instruction is supported.
//!
//! # Examples
//!
//! ```
//! # use std::error::Error;
//! #
//! # fn main() -> Result<(), Box<dyn Error>> {
//! use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode, Mnemonic, OperandSize};
//!
//! // `VMXON qword ptr [rax]`
//! let ins = DecodedInstruction::decode(&[0xf3, 0x0f, 0xc7, 0x30], DecodeMode::Bits64)?;
//! let modes = ins.valid_cpu_modes();
//!
//! // Check if the instruction is available from user mode
//! if modes.privilege_level.ring3 {
//! println!("Available in user mode");
//! } else {
//! println!("Not available in user mode");
//! }
//!
//! # Ok(())
//! # }
extern crate bddisasm_sys as ffi;
// TODO: maybe use something like the `bitflags` crate and have all these as flags?
/// Privilege levels (rings) in which an instruction is supported.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct PrivilegeLevel {
pub ring0: bool,
pub ring1: bool,
pub ring2: bool,
pub ring3: bool,
}
/// Operating modes in which an instruction is supported.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct OperatingMode {
/// The instruction is valid in real mode.
pub real: bool,
/// The instruction is valid in Virtual 8086 mode.
pub v8086: bool,
/// The instruction is valid in protected mode (32 bit).
pub protected: bool,
/// The instruction is valid in compatibility mode (32 bit in 64 bit).
pub compat: bool,
/// The instruction is valid in long mode.
pub long: bool,
}
/// Special modes - these may be active inside other modes (example: `TSX` in `Long mode`).
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct SpecialModes {
/// The instruction is valid in System Management Mode.
pub smm: bool,
/// The instruction is valid outside System Management Mode.
pub smm_off: bool,
/// The instruction is valid in SGX mode.
pub sgx: bool,
/// The instruction is valid outside SGX mode.
pub sgx_off: bool,
/// The instruction is valid in transactional regions.
pub tsx: bool,
/// The instruction is valid in transactional regions.
pub tsx_off: bool,
}
/// VMX mode - they engulf all the other modes.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct VmxMode {
/// The instruction is valid in VMX root mode.
pub root: bool,
/// The instruction is valid in VMX non root mode.
pub non_root: bool,
/// The instruction is valid in VMX root SEAM.
pub root_seam: bool,
/// The instruction is valid in VMX non-root SEAM.
pub non_root_seam: bool,
/// The instruction is valid outside VMX operation.
pub off: bool,
}
/// Indicates in which modes the instruction is valid in
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct CpuModes {
pub privilege_level: PrivilegeLevel,
pub operating_mode: OperatingMode,
pub special_modes: SpecialModes,
pub vmx: VmxMode,
}
#[doc(hidden)]
impl From<ffi::ND_VALID_MODES> for CpuModes {
fn from(raw: ffi::ND_VALID_MODES) -> CpuModes {
let raw = unsafe { raw.__bindgen_anon_1 };
CpuModes {
privilege_level: PrivilegeLevel {
ring0: raw.Ring0() != 0,
ring1: raw.Ring1() != 0,
ring2: raw.Ring2() != 0,
ring3: raw.Ring3() != 0,
},
operating_mode: OperatingMode {
real: raw.Real() != 0,
v8086: raw.V8086() != 0,
protected: raw.Protected() != 0,
compat: raw.Compat() != 0,
long: raw.Long() != 0,
},
special_modes: SpecialModes {
smm: raw.Smm() != 0,
smm_off: raw.SmmOff() != 0,
sgx: raw.Sgx() != 0,
sgx_off: raw.SgxOff() != 0,
tsx: raw.Tsx() != 0,
tsx_off: raw.TsxOff() != 0,
},
vmx: VmxMode {
root: raw.VmxRoot() != 0,
non_root: raw.VmxNonRoot() != 0,
root_seam: raw.VmxRootSeam() != 0,
non_root_seam: raw.VmxNonRootSeam() != 0,
off: raw.VmxOff() != 0,
},
}
}
}

@ -0,0 +1,63 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! CPUID information used to check if an instruction is supported on a CPU or not.
//!
//! # Examples
//!
//! ```
//! # use std::error::Error;
//! #
//! # fn test() -> Option<()> {
//! use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode};
//!
//! // `ENCLS`
//! let ins = DecodedInstruction::decode(&[0x0f, 0x01, 0xcf], DecodeMode::Bits64).ok()?;
//! let cpuid = ins.cpuid()?;
//! println!("Leaf: {}", cpuid.leaf);
//! match cpuid.sub_leaf {
//! Some(sub_leaf) => println!("Sub-leaf: {}", sub_leaf),
//! None => println!("Sub-leaf: None"),
//! }
//! println!("Register to check: {} Bit: {}", cpuid.register, cpuid.bit);
//!
//! # Some(())
//! # }
extern crate bddisasm_sys as ffi;
use std::fmt;
/// Describes the CPUID leaf, sub-leaf, register & bit that indicate whether an instruction is supported or not.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Cpuid {
/// CPUID leaf.
pub leaf: u32,
/// CPUID sub-leaf.
pub sub_leaf: Option<u32>,
/// The register that contains information regarding the instruction.
pub register: u8,
/// Bit inside the register that indicates whether the instruction is present.
pub bit: u64,
}
impl fmt::Display for Cpuid {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self.sub_leaf {
Some(sub_leaf) => write!(
f,
"leaf: {:#x} sub-leaf: {:#x} register: {} bit to test: {:#x}",
self.leaf, sub_leaf, self.register, self.bit
),
None => write!(
f,
"leaf: {:#x} sub-leaf: - register: {} bit to test: {:#x}",
self.leaf, self.register, self.bit
),
}
}
}

@ -0,0 +1,256 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Errors that can be encountered when decoding an instruction or when trying to get details about a decoded
//! instruction.
//!
//! # Notes
//!
//! All error codes that can be returned by `bddisasm-sys` are encapsulated in the [DecodeError](DecodeError) enum.
//! However, some of these are unlikely to be encountered when using this crate (for example,
//! [BufferOverflow](DecodeError::BufferOverflow)) which indicates that a buffer passed to the `bddisasm` C library is
//! not large enough.
//!
//! Other errors, such as [UnknownStatus](DecodeError::UnknownStatus) or
//! [UnknownInstruction](DecodeError::UnknownInstruction) are used to indicate that this crate is out of sync with
//! `bddisasm-sys`, which also should never happen.
extern crate bddisasm_sys as ffi;
use std::error::Error;
use std::fmt;
/// Holds all the possible errors that can be encountered by the decoder.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum DecodeError {
/// The underlying bddisasm status is not known. The inner value holds the status as it was returned by bddisasm.
///
/// This usually means that this library is out of sync with bddisasm-sys and is not aware that a new error status
/// was added.
UnknownStatus(u32),
/// The instruction class returned by bddisasm is not known. The inner value holds the instruction class as it was
/// returned by bddisasm.
///
/// This usually means that this library is out of sync with bddisasm-sys and is not aware that a new instruction
/// class was added.
UnknownInstruction(u32),
/// The provided input buffer is too small and does not contain a valid instruction.
BufferTooSmall,
/// Invalid encoding/instruction.
InvalidEncoding,
/// Instruction exceeds the maximum 15 bytes.
InstructionTooLong,
/// Invalid prefix sequence is present.
InvalidPrefixSequence,
/// The instruction uses an invalid register.
InvalidRegisterInInstruction,
/// XOP is present, but also a legacy prefix.
XopWithPrefix,
/// VEX is present, but also a legacy prefix.
VexWithPrefix,
/// EVEX is present, but also a legacy prefix.
EvexWithPrefix,
/// Invalid encoding/instruction.
InvalidEncodingInMode,
/// Invalid usage of LOCK.
BadLockPrefix,
/// An attempt to load the CS register.
CsLoad,
/// 0x66 prefix is not accepted.
Prefix66NotAccepted,
/// 16 bit addressing mode not supported.
AddressingNotSupported16Bit,
/// RIP-relative addressing not supported.
RipRelAddressingNotSupported,
/// Instruction uses VSIB, but SIB is not present.
VsibWithoutSib,
/// VSIB addressing, same vector reg used more than once.
InvalidVsibRegs,
/// VEX.VVVV field must be zero.
VexVvvvMustBeZero,
/// Masking is not supported.
MaskNotSupported,
/// Masking is mandatory.
MaskRequired,
/// Embedded rounding/SAE not supported.
ErSaeNotSupported,
/// Zeroing not supported.
ZeroingNotSupported,
/// Zeroing on memory.
ZeroingOnMemory,
/// Zeroing without masking.
ZeroingNoMask,
/// Broadcast not supported.
BroadcastNotSupported,
/// EVEX.V' field must be one (negated 0).
BadEvexVPrime,
/// EVEX.L'L field is invalid for the instruction.
BadEvexLl,
/// Instruction uses SIBMEM, but SIB is not present.
SibmemWithoutSib,
/// Tile registers are not unique.
InvalidTileRegs,
/// Destination register is not unique (used as src).
InvalidDestRegs,
/// An invalid parameter was provided.
InvalidParameter,
/// The INSTRUX contains unexpected values.
InvalidInstrux,
/// Not enough space is available.
BufferOverflow,
/// Internal library error.
InternalError,
}
impl fmt::Display for DecodeError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self {
DecodeError::UnknownStatus(value) => write!(f, "unknown status: {:#x}", value),
DecodeError::UnknownInstruction(value) => {
write!(f, "unknown instruction: {:#x}", value)
}
DecodeError::BufferTooSmall => write!(f, "the provided input buffer is too small"),
DecodeError::InvalidEncoding => write!(f, "invalid encoding/instruction"),
DecodeError::InstructionTooLong => {
write!(f, "instruction exceeds the maximum 15 bytes")
}
DecodeError::InvalidPrefixSequence => write!(f, "invalid prefix sequence is present"),
DecodeError::InvalidRegisterInInstruction => {
write!(f, "the instruction uses an invalid register")
}
DecodeError::XopWithPrefix => write!(f, "XOP is present, but also a legacy prefix"),
DecodeError::VexWithPrefix => write!(f, "VEX is present, but also a legacy prefix"),
DecodeError::EvexWithPrefix => write!(f, "EVEX is present, but also a legacy prefix"),
DecodeError::InvalidEncodingInMode => write!(f, "invalid encoding/instruction"),
DecodeError::BadLockPrefix => write!(f, "invalid usage of LOCK"),
DecodeError::CsLoad => write!(f, "an attempt to load the CS register"),
DecodeError::Prefix66NotAccepted => write!(f, "0x66 prefix is not accepted"),
DecodeError::AddressingNotSupported16Bit => {
write!(f, "16 bit addressing mode not supported")
}
DecodeError::RipRelAddressingNotSupported => {
write!(f, "RIP-relative addressing not supported")
}
DecodeError::VsibWithoutSib => {
write!(f, "instruction uses VSIB, but SIB is not present")
}
DecodeError::InvalidVsibRegs => {
write!(
f,
"VSIB addressing with the same vector register used more than once"
)
}
DecodeError::VexVvvvMustBeZero => write!(f, "VEX.VVVV field must be zero"),
DecodeError::MaskNotSupported => write!(f, "masking is not supported"),
DecodeError::MaskRequired => write!(f, "masking is mandatory"),
DecodeError::ErSaeNotSupported => write!(f, "embedded rounding/SAE not supported"),
DecodeError::ZeroingNotSupported => write!(f, "zeroing not supported"),
DecodeError::ZeroingOnMemory => write!(f, "zeroing on memory"),
DecodeError::ZeroingNoMask => write!(f, "zeroing without masking"),
DecodeError::BroadcastNotSupported => write!(f, "broadcast not supported"),
DecodeError::BadEvexVPrime => write!(f, "EVEX.V' field must be one (negated 0)"),
DecodeError::BadEvexLl => write!(f, "EVEX.L'L field is invalid for the instruction"),
DecodeError::SibmemWithoutSib => {
write!(f, "instruction uses SIBMEM, but SIB is not present")
}
DecodeError::InvalidTileRegs => write!(f, "tile registers are not unique"),
DecodeError::InvalidDestRegs => {
write!(f, "destination register is not unique (used as src)")
}
DecodeError::InvalidParameter => write!(f, "an invalid parameter was provided"),
DecodeError::InvalidInstrux => {
write!(f, "the INSTRUX structure contains unexpected values")
}
DecodeError::BufferOverflow => {
write!(f, "not enough space is available to format instruction")
}
DecodeError::InternalError => write!(f, "internal error"),
}
}
}
impl Error for DecodeError {}
pub(crate) fn status_to_error(status: ffi::NDSTATUS) -> Result<(), DecodeError> {
if status == ffi::ND_STATUS_SUCCESS || status == ffi::ND_STATUS_HINT_OPERAND_NOT_USED {
Ok(())
} else {
match status {
ffi::ND_STATUS_BUFFER_TOO_SMALL => Err(DecodeError::BufferTooSmall),
ffi::ND_STATUS_INVALID_ENCODING => Err(DecodeError::InvalidEncoding),
ffi::ND_STATUS_INSTRUCTION_TOO_LONG => Err(DecodeError::InstructionTooLong),
ffi::ND_STATUS_INVALID_PREFIX_SEQUENCE => Err(DecodeError::InvalidPrefixSequence),
ffi::ND_STATUS_INVALID_REGISTER_IN_INSTRUCTION => {
Err(DecodeError::InvalidRegisterInInstruction)
}
ffi::ND_STATUS_XOP_WITH_PREFIX => Err(DecodeError::XopWithPrefix),
ffi::ND_STATUS_VEX_WITH_PREFIX => Err(DecodeError::VexWithPrefix),
ffi::ND_STATUS_EVEX_WITH_PREFIX => Err(DecodeError::EvexWithPrefix),
ffi::ND_STATUS_INVALID_ENCODING_IN_MODE => Err(DecodeError::InvalidEncodingInMode),
ffi::ND_STATUS_BAD_LOCK_PREFIX => Err(DecodeError::BadLockPrefix),
ffi::ND_STATUS_CS_LOAD => Err(DecodeError::CsLoad),
ffi::ND_STATUS_66_NOT_ACCEPTED => Err(DecodeError::Prefix66NotAccepted),
ffi::ND_STATUS_16_BIT_ADDRESSING_NOT_SUPPORTED => {
Err(DecodeError::AddressingNotSupported16Bit)
}
ffi::ND_STATUS_RIP_REL_ADDRESSING_NOT_SUPPORTED => {
Err(DecodeError::RipRelAddressingNotSupported)
}
ffi::ND_STATUS_VSIB_WITHOUT_SIB => Err(DecodeError::VsibWithoutSib),
ffi::ND_STATUS_INVALID_VSIB_REGS => Err(DecodeError::InvalidVsibRegs),
ffi::ND_STATUS_VEX_VVVV_MUST_BE_ZERO => Err(DecodeError::VexVvvvMustBeZero),
ffi::ND_STATUS_MASK_NOT_SUPPORTED => Err(DecodeError::MaskNotSupported),
ffi::ND_STATUS_MASK_REQUIRED => Err(DecodeError::MaskRequired),
ffi::ND_STATUS_ER_SAE_NOT_SUPPORTED => Err(DecodeError::ErSaeNotSupported),
ffi::ND_STATUS_ZEROING_NOT_SUPPORTED => Err(DecodeError::ZeroingNotSupported),
ffi::ND_STATUS_ZEROING_ON_MEMORY => Err(DecodeError::ZeroingOnMemory),
ffi::ND_STATUS_ZEROING_NO_MASK => Err(DecodeError::ZeroingNoMask),
ffi::ND_STATUS_BROADCAST_NOT_SUPPORTED => Err(DecodeError::BroadcastNotSupported),
ffi::ND_STATUS_BAD_EVEX_V_PRIME => Err(DecodeError::BadEvexVPrime),
ffi::ND_STATUS_BAD_EVEX_LL => Err(DecodeError::BadEvexLl),
ffi::ND_STATUS_SIBMEM_WITHOUT_SIB => Err(DecodeError::SibmemWithoutSib),
ffi::ND_STATUS_INVALID_TILE_REGS => Err(DecodeError::InvalidTileRegs),
ffi::ND_STATUS_INVALID_DEST_REGS => Err(DecodeError::InvalidDestRegs),
ffi::ND_STATUS_INVALID_PARAMETER => Err(DecodeError::InvalidParameter),
ffi::ND_STATUS_INVALID_INSTRUX => Err(DecodeError::InvalidInstrux),
ffi::ND_STATUS_BUFFER_OVERFLOW => Err(DecodeError::BufferOverflow),
ffi::ND_STATUS_INTERNAL_ERROR => Err(DecodeError::InternalError),
_ => Err(DecodeError::UnknownStatus(status)),
}
}
}

File diff suppressed because it is too large Load Diff

@ -0,0 +1,215 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Decodes instructions.
extern crate bddisasm_sys as ffi;
pub use crate::decode_error::DecodeError;
pub use crate::decoded_instruction::{DecodeMode, DecodeResult, DecodedInstruction};
pub use crate::mnemonic::Mnemonic;
/// Decodes instructions.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Decoder<'a> {
code: &'a [u8],
ip: u64,
mode: DecodeMode,
offset: usize,
}
impl<'a> Decoder<'a> {
/// Creates a new decoder.
///
/// # Arguments
///
/// * `code` - An [u8](u8) slice that holds the code to be decoded.
/// * `mode` - The mode in which to decode the instruction.
/// * `ip` - The instruction pointer value to use when formatting the decoded instruction. Does not affect the
/// decoding process in any way.
pub fn new(code: &'a [u8], mode: DecodeMode, ip: u64) -> Self {
Self {
code,
mode,
ip,
offset: 0,
}
}
/// Attempts to decode the next instruction from the given code chunk.
///
/// # Returns
///
/// * `Some(DecodeResult)` - if there are still undecoded bytes in the given code chunk. The decoding may have
/// still failed. See `Remarks`.
/// * `None` - if all the bytes in the given code chunk were decoded.
///
/// # Remarks
///
/// This function decodes one instruction from the given code chunk at a time. After each call, the offset inside
/// the code chunk is advanced by:
///
/// - the size of the decoded instruction, if decoding was succesfull
/// - 1, if decoding was not succesfull
///
/// The `ip` value specified when the decoder was created is automatically updated:
///
/// - if the decoding was succesfull, it is incremented with the instruction size
/// - if the decoding was not succesfull, it is incremented with 1
///
/// # Examples
///
/// ```
/// # use std::error::Error;
/// #
/// # fn main() -> Result<(), Box<dyn Error>> {
/// use bddisasm::decoder::{Decoder, DecodeMode};
///
/// let mut decoder = Decoder::new(&[0x51, 0x53], DecodeMode::Bits32, 0);
///
/// // As long as we have something to decode
/// while let Some(result) = decoder.decode_next() {
/// // Check if the decoding succeeded
/// match result {
/// Ok(instruction) => println!("{}", instruction),
/// Err(e) => println!("Unable to decode: {}", e),
/// }
/// }
///
/// # Ok(())
/// # }
/// ```
pub fn decode_next(&mut self) -> Option<DecodeResult> {
if self.offset >= self.code.len() {
None
} else {
let result =
DecodedInstruction::decode_with_ip(&self.code[self.offset..], self.mode, self.ip);
match result {
Ok(ins) => {
self.offset += ins.length() as usize;
self.ip += ins.length() as u64;
}
Err(_) => {
self.offset += 1;
self.ip += 1;
}
};
Some(result)
}
}
/// Attempts to decode the next instruction from the given code chunk.
///
/// Behaves like [decode_next](Decoder::decode_next), but in addition to the [DecodeResult](DecodeResult) it will
/// also return the offset from which decoding was attempted, as well as the corresponding instruction pointer.
///
/// # Examples
///
/// ```
/// # use std::error::Error;
/// #
/// # fn main() -> Result<(), Box<dyn Error>> {
/// use bddisasm::decoder::{Decoder, DecodeMode};
///
/// let mut decoder = Decoder::new(&[0x51, 0x53], DecodeMode::Bits32, 0x1000);
///
/// // As long as we have something to decode
/// while let Some((result, offset, ip)) = decoder.decode_next_with_info() {
/// // Check if the decoding succeeded
/// match result {
/// Ok(instruction) => println!("IP: {:#x} {}", ip, instruction),
/// Err(e) => println!("Unable to decode at offset {:#x}: {}", offset, e),
/// }
/// }
///
/// # Ok(())
/// # }
/// ```
pub fn decode_next_with_info(&mut self) -> Option<(DecodeResult, usize, u64)> {
let offset = self.offset;
let ip = self.ip;
self.decode_next().map(|res| (res, offset, ip))
}
}
impl Iterator for Decoder<'_> {
type Item = DecodeResult;
#[inline]
fn next(&mut self) -> Option<Self::Item> {
self.decode_next()
}
}
impl std::iter::FusedIterator for Decoder<'_> {}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn decode_next() {
let code = vec![0xb8, 0x00, 0x00, 0x00, 0x00, 0x48, 0x8b, 0xf9, 0xff, 0xff];
let mut decoder = Decoder::new(&code, DecodeMode::Bits64, 0x1000);
let expected: Vec<Result<(Mnemonic, &str, &[u8]), DecodeError>> = vec![
Ok((
Mnemonic::Mov,
"MOV eax, 0x00000000",
&[0xb8, 0x00, 0x00, 0x00, 0x00],
)),
Ok((Mnemonic::Mov, "MOV rdi, rcx", &[0x48, 0x8b, 0xf9])),
Err(DecodeError::InvalidEncoding),
Err(DecodeError::BufferTooSmall),
];
let mut exected_index = 0usize;
while let Some(ins) = decoder.decode_next() {
match expected[exected_index] {
Ok((i, s, b)) => {
let ins = ins.expect("Unable to decode");
assert_eq!(i, ins.mnemonic());
assert_eq!(b, ins.bytes());
assert_eq!(s, format!("{}", ins));
}
Err(e) => {
assert_eq!(e, ins.expect_err("Expected error"));
}
};
exected_index += 1;
}
}
#[test]
fn decoder_iter() {
let code = vec![0xb8, 0x00, 0x00, 0x00, 0x00, 0x48, 0x8b, 0xf9, 0xff, 0xff];
let decoder = Decoder::new(&code, DecodeMode::Bits64, 0x1000);
let expected: Vec<Result<(Mnemonic, &str, &[u8]), DecodeError>> = vec![
Ok((
Mnemonic::Mov,
"MOV eax, 0x00000000",
&[0xb8, 0x00, 0x00, 0x00, 0x00],
)),
Ok((Mnemonic::Mov, "MOV rdi, rcx", &[0x48, 0x8b, 0xf9])),
Err(DecodeError::InvalidEncoding),
Err(DecodeError::BufferTooSmall),
];
for (index, ins) in decoder.enumerate() {
match expected[index] {
Ok((i, s, b)) => {
let ins = ins.expect("Unable to decode");
assert_eq!(i, ins.mnemonic());
assert_eq!(b, ins.bytes());
assert_eq!(s, format!("{}", ins));
}
Err(e) => {
assert_eq!(e, ins.expect_err("Expected error"));
}
};
}
}
}

@ -0,0 +1,58 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Offers information about how an instructions accesses the FPU status registers.
extern crate bddisasm_sys as ffi;
/// The mode in which a FPU status flag is accessed.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum FpuFlagsAccess {
/// The FPU status flag is cleared to 0.
Cleared,
/// The FPU status flag is set to 1.
Set,
/// The FPU status flag is modified according to a result.
Modified,
/// The FPU status flag is undefined or unaffected.
Undefined,
}
impl From<u8> for FpuFlagsAccess {
fn from(value: u8) -> FpuFlagsAccess {
let value = value as u32;
match value {
ffi::ND_FPU_FLAG_SET_0 => FpuFlagsAccess::Cleared,
ffi::ND_FPU_FLAG_SET_1 => FpuFlagsAccess::Set,
ffi::ND_FPU_FLAG_MODIFIED => FpuFlagsAccess::Modified,
ffi::ND_FPU_FLAG_UNDEFINED => FpuFlagsAccess::Undefined,
_ => panic!("Unexpected FPU flag access: {}", value),
}
}
}
/// Describes the way an instruction accesses the FPU flags registers.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct FpuFlags {
/// C0 flag access mode.
pub c0: FpuFlagsAccess,
/// C1 flag access mode.
pub c1: FpuFlagsAccess,
/// C2 flag access mode.
pub c2: FpuFlagsAccess,
/// C3 flag access mode.
pub c3: FpuFlagsAccess,
}
#[doc(hidden)]
impl From<ffi::ND_FPU_FLAGS> for FpuFlags {
fn from(flags: ffi::ND_FPU_FLAGS) -> FpuFlags {
FpuFlags {
c0: FpuFlagsAccess::from(flags.C0()),
c1: FpuFlagsAccess::from(flags.C1()),
c2: FpuFlagsAccess::from(flags.C2()),
c3: FpuFlagsAccess::from(flags.C3()),
}
}
}

@ -0,0 +1,236 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Instruction categories.
extern crate bddisasm_sys as ffi;
use super::decode_error;
use std::convert::TryFrom;
/// Instruction category.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum Category {
I3dnow,
Aes,
Aeskl,
Amx,
Arith,
Avx,
Avx2,
Avx2gather,
Avx512,
Avx512bf16,
Avx512fp16,
Avx512vbmi,
Avx512vp2intersect,
Avxvnni,
Bitbyte,
Blend,
Bmi1,
Bmi2,
Broadcast,
Call,
Cet,
Cldemote,
Cmov,
Compress,
CondBr,
Conflict,
Convert,
Dataxfer,
Decimal,
Enqcmd,
Expand,
Flagop,
Fma4,
Fred,
Gather,
Gfni,
Hreset,
I386,
Ifma,
Interrupt,
Io,
Iostringop,
Kl,
Kmask,
Knl,
Lkgs,
Logic,
Logical,
LogicalFp,
Lwp,
Lzcnt,
Misc,
Mmx,
Movdir64b,
Movdiri,
Mpx,
Nop,
Padlock,
Pclmulqdq,
Pconfig,
Pop,
Prefetch,
Ptwrite,
Push,
Rdpid,
Rdrand,
Rdseed,
Rdwrfsgs,
Ret,
Rotate,
Scatter,
Segop,
Semaphore,
Sgx,
Sha,
Shift,
Smap,
Sse,
Sse2,
Stringop,
Sttni,
Syscall,
Sysret,
System,
Tdx,
Ud,
Uintr,
UncondBr,
Unknown,
Vaes,
Vfma,
Vfmaps,
Vnni,
Vnniw,
Vpclmulqdq,
Vpopcnt,
Vtx,
Waitpkg,
Wbnoinvd,
Widenop,
WideKl,
X87Alu,
Xop,
Xsave,
}
#[doc(hidden)]
impl TryFrom<ffi::ND_INS_CATEGORY> for Category {
type Error = decode_error::DecodeError;
fn try_from(value: ffi::ND_INS_CATEGORY) -> Result<Self, Self::Error> {
match value {
ffi::_ND_INS_TYPE::ND_CAT_INVALID => {
Err(decode_error::DecodeError::UnknownInstruction(value as u32))
}
ffi::_ND_INS_TYPE::ND_CAT_3DNOW => Ok(Category::I3dnow),
ffi::_ND_INS_TYPE::ND_CAT_AES => Ok(Category::Aes),
ffi::_ND_INS_TYPE::ND_CAT_AESKL => Ok(Category::Aeskl),
ffi::_ND_INS_TYPE::ND_CAT_AMX => Ok(Category::Amx),
ffi::_ND_INS_TYPE::ND_CAT_ARITH => Ok(Category::Arith),
ffi::_ND_INS_TYPE::ND_CAT_AVX => Ok(Category::Avx),
ffi::_ND_INS_TYPE::ND_CAT_AVX2 => Ok(Category::Avx2),
ffi::_ND_INS_TYPE::ND_CAT_AVX2GATHER => Ok(Category::Avx2gather),
ffi::_ND_INS_TYPE::ND_CAT_AVX512 => Ok(Category::Avx512),
ffi::_ND_INS_TYPE::ND_CAT_AVX512BF16 => Ok(Category::Avx512bf16),
ffi::_ND_INS_TYPE::ND_CAT_AVX512FP16 => Ok(Category::Avx512fp16),
ffi::_ND_INS_TYPE::ND_CAT_AVX512VBMI => Ok(Category::Avx512vbmi),
ffi::_ND_INS_TYPE::ND_CAT_AVX512VP2INTERSECT => Ok(Category::Avx512vp2intersect),
ffi::_ND_INS_TYPE::ND_CAT_AVXVNNI => Ok(Category::Avxvnni),
ffi::_ND_INS_TYPE::ND_CAT_BITBYTE => Ok(Category::Bitbyte),
ffi::_ND_INS_TYPE::ND_CAT_BLEND => Ok(Category::Blend),
ffi::_ND_INS_TYPE::ND_CAT_BMI1 => Ok(Category::Bmi1),
ffi::_ND_INS_TYPE::ND_CAT_BMI2 => Ok(Category::Bmi2),
ffi::_ND_INS_TYPE::ND_CAT_BROADCAST => Ok(Category::Broadcast),
ffi::_ND_INS_TYPE::ND_CAT_CALL => Ok(Category::Call),
ffi::_ND_INS_TYPE::ND_CAT_CET => Ok(Category::Cet),
ffi::_ND_INS_TYPE::ND_CAT_CLDEMOTE => Ok(Category::Cldemote),
ffi::_ND_INS_TYPE::ND_CAT_CMOV => Ok(Category::Cmov),
ffi::_ND_INS_TYPE::ND_CAT_COMPRESS => Ok(Category::Compress),
ffi::_ND_INS_TYPE::ND_CAT_COND_BR => Ok(Category::CondBr),
ffi::_ND_INS_TYPE::ND_CAT_CONFLICT => Ok(Category::Conflict),
ffi::_ND_INS_TYPE::ND_CAT_CONVERT => Ok(Category::Convert),
ffi::_ND_INS_TYPE::ND_CAT_DATAXFER => Ok(Category::Dataxfer),
ffi::_ND_INS_TYPE::ND_CAT_DECIMAL => Ok(Category::Decimal),
ffi::_ND_INS_TYPE::ND_CAT_ENQCMD => Ok(Category::Enqcmd),
ffi::_ND_INS_TYPE::ND_CAT_EXPAND => Ok(Category::Expand),
ffi::_ND_INS_TYPE::ND_CAT_FLAGOP => Ok(Category::Flagop),
ffi::_ND_INS_TYPE::ND_CAT_FMA4 => Ok(Category::Fma4),
ffi::_ND_INS_TYPE::ND_CAT_FRED => Ok(Category::Fred),
ffi::_ND_INS_TYPE::ND_CAT_GATHER => Ok(Category::Gather),
ffi::_ND_INS_TYPE::ND_CAT_GFNI => Ok(Category::Gfni),
ffi::_ND_INS_TYPE::ND_CAT_HRESET => Ok(Category::Hreset),
ffi::_ND_INS_TYPE::ND_CAT_I386 => Ok(Category::I386),
ffi::_ND_INS_TYPE::ND_CAT_IFMA => Ok(Category::Ifma),
ffi::_ND_INS_TYPE::ND_CAT_INTERRUPT => Ok(Category::Interrupt),
ffi::_ND_INS_TYPE::ND_CAT_IO => Ok(Category::Io),
ffi::_ND_INS_TYPE::ND_CAT_IOSTRINGOP => Ok(Category::Iostringop),
ffi::_ND_INS_TYPE::ND_CAT_KL => Ok(Category::Kl),
ffi::_ND_INS_TYPE::ND_CAT_KMASK => Ok(Category::Kmask),
ffi::_ND_INS_TYPE::ND_CAT_KNL => Ok(Category::Knl),
ffi::_ND_INS_TYPE::ND_CAT_LKGS => Ok(Category::Lkgs),
ffi::_ND_INS_TYPE::ND_CAT_LOGIC => Ok(Category::Logic),
ffi::_ND_INS_TYPE::ND_CAT_LOGICAL => Ok(Category::Logical),
ffi::_ND_INS_TYPE::ND_CAT_LOGICAL_FP => Ok(Category::LogicalFp),
ffi::_ND_INS_TYPE::ND_CAT_LWP => Ok(Category::Lwp),
ffi::_ND_INS_TYPE::ND_CAT_LZCNT => Ok(Category::Lzcnt),
ffi::_ND_INS_TYPE::ND_CAT_MISC => Ok(Category::Misc),
ffi::_ND_INS_TYPE::ND_CAT_MMX => Ok(Category::Mmx),
ffi::_ND_INS_TYPE::ND_CAT_MOVDIR64B => Ok(Category::Movdir64b),
ffi::_ND_INS_TYPE::ND_CAT_MOVDIRI => Ok(Category::Movdiri),
ffi::_ND_INS_TYPE::ND_CAT_MPX => Ok(Category::Mpx),
ffi::_ND_INS_TYPE::ND_CAT_NOP => Ok(Category::Nop),
ffi::_ND_INS_TYPE::ND_CAT_PADLOCK => Ok(Category::Padlock),
ffi::_ND_INS_TYPE::ND_CAT_PCLMULQDQ => Ok(Category::Pclmulqdq),
ffi::_ND_INS_TYPE::ND_CAT_PCONFIG => Ok(Category::Pconfig),
ffi::_ND_INS_TYPE::ND_CAT_POP => Ok(Category::Pop),
ffi::_ND_INS_TYPE::ND_CAT_PREFETCH => Ok(Category::Prefetch),
ffi::_ND_INS_TYPE::ND_CAT_PTWRITE => Ok(Category::Ptwrite),
ffi::_ND_INS_TYPE::ND_CAT_PUSH => Ok(Category::Push),
ffi::_ND_INS_TYPE::ND_CAT_RDPID => Ok(Category::Rdpid),
ffi::_ND_INS_TYPE::ND_CAT_RDRAND => Ok(Category::Rdrand),
ffi::_ND_INS_TYPE::ND_CAT_RDSEED => Ok(Category::Rdseed),
ffi::_ND_INS_TYPE::ND_CAT_RDWRFSGS => Ok(Category::Rdwrfsgs),
ffi::_ND_INS_TYPE::ND_CAT_RET => Ok(Category::Ret),
ffi::_ND_INS_TYPE::ND_CAT_ROTATE => Ok(Category::Rotate),
ffi::_ND_INS_TYPE::ND_CAT_SCATTER => Ok(Category::Scatter),
ffi::_ND_INS_TYPE::ND_CAT_SEGOP => Ok(Category::Segop),
ffi::_ND_INS_TYPE::ND_CAT_SEMAPHORE => Ok(Category::Semaphore),
ffi::_ND_INS_TYPE::ND_CAT_SGX => Ok(Category::Sgx),
ffi::_ND_INS_TYPE::ND_CAT_SHA => Ok(Category::Sha),
ffi::_ND_INS_TYPE::ND_CAT_SHIFT => Ok(Category::Shift),
ffi::_ND_INS_TYPE::ND_CAT_SMAP => Ok(Category::Smap),
ffi::_ND_INS_TYPE::ND_CAT_SSE => Ok(Category::Sse),
ffi::_ND_INS_TYPE::ND_CAT_SSE2 => Ok(Category::Sse2),
ffi::_ND_INS_TYPE::ND_CAT_STRINGOP => Ok(Category::Stringop),
ffi::_ND_INS_TYPE::ND_CAT_STTNI => Ok(Category::Sttni),
ffi::_ND_INS_TYPE::ND_CAT_SYSCALL => Ok(Category::Syscall),
ffi::_ND_INS_TYPE::ND_CAT_SYSRET => Ok(Category::Sysret),
ffi::_ND_INS_TYPE::ND_CAT_SYSTEM => Ok(Category::System),
ffi::_ND_INS_TYPE::ND_CAT_TDX => Ok(Category::Tdx),
ffi::_ND_INS_TYPE::ND_CAT_UD => Ok(Category::Ud),
ffi::_ND_INS_TYPE::ND_CAT_UINTR => Ok(Category::Uintr),
ffi::_ND_INS_TYPE::ND_CAT_UNCOND_BR => Ok(Category::UncondBr),
ffi::_ND_INS_TYPE::ND_CAT_UNKNOWN => Ok(Category::Unknown),
ffi::_ND_INS_TYPE::ND_CAT_VAES => Ok(Category::Vaes),
ffi::_ND_INS_TYPE::ND_CAT_VFMA => Ok(Category::Vfma),
ffi::_ND_INS_TYPE::ND_CAT_VFMAPS => Ok(Category::Vfmaps),
ffi::_ND_INS_TYPE::ND_CAT_VNNI => Ok(Category::Vnni),
ffi::_ND_INS_TYPE::ND_CAT_VNNIW => Ok(Category::Vnniw),
ffi::_ND_INS_TYPE::ND_CAT_VPCLMULQDQ => Ok(Category::Vpclmulqdq),
ffi::_ND_INS_TYPE::ND_CAT_VPOPCNT => Ok(Category::Vpopcnt),
ffi::_ND_INS_TYPE::ND_CAT_VTX => Ok(Category::Vtx),
ffi::_ND_INS_TYPE::ND_CAT_WAITPKG => Ok(Category::Waitpkg),
ffi::_ND_INS_TYPE::ND_CAT_WBNOINVD => Ok(Category::Wbnoinvd),
ffi::_ND_INS_TYPE::ND_CAT_WIDENOP => Ok(Category::Widenop),
ffi::_ND_INS_TYPE::ND_CAT_WIDE_KL => Ok(Category::WideKl),
ffi::_ND_INS_TYPE::ND_CAT_X87_ALU => Ok(Category::X87Alu),
ffi::_ND_INS_TYPE::ND_CAT_XOP => Ok(Category::Xop),
ffi::_ND_INS_TYPE::ND_CAT_XSAVE => Ok(Category::Xsave),
}
}
}

@ -0,0 +1,262 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Instruction sets.
extern crate bddisasm_sys as ffi;
use super::decode_error;
use std::convert::TryFrom;
/// ISA set.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum IsaSet {
I3dnow,
Adx,
Aes,
Amd,
Amxbf16,
Amxint8,
Amxtile,
Avx,
Avx2,
Avx2gather,
Avx5124fmaps,
Avx5124vnniw,
Avx512bf16,
Avx512bitalg,
Avx512bw,
Avx512cd,
Avx512dq,
Avx512er,
Avx512f,
Avx512fp16,
Avx512ifma,
Avx512pf,
Avx512vbmi,
Avx512vbmi2,
Avx512vnni,
Avx512vp2intersect,
Avx512vpopcntdq,
Avxvnni,
Bmi1,
Bmi2,
CetIbt,
CetSs,
Cldemote,
Clfsh,
Clfshopt,
Clwb,
Clzero,
Cmpxchg16b,
Cyrix,
CyrixSmm,
Enqcmd,
F16c,
Fma,
Fma4,
Fred,
Fxsave,
Gfni,
Hreset,
I186,
I286prot,
I286real,
I386,
I486,
I486real,
I64,
I86,
Invlpgb,
Invpcid,
Kl,
Lkgs,
Longmode,
Lwp,
Lzcnt,
Mcommit,
Mmx,
Movbe,
Movdir64b,
Movdiri,
Mpx,
Mwaitt,
Pause,
Pclmulqdq,
Pconfig,
Pentiumreal,
Pku,
Popcnt,
Ppro,
PrefetchNop,
Ptwrite,
Rdpid,
Rdpmc,
Rdpru,
Rdrand,
Rdseed,
Rdtscp,
Rdwrfsgs,
Serialize,
Sgx,
Sha,
Smap,
Smx,
Snp,
Sse,
Sse2,
Sse3,
Sse4,
Sse42,
Sse4a,
Ssse3,
Svm,
Tbm,
Tdx,
Tsx,
Tsxldtrk,
Ud,
Uintr,
Unknown,
Vaes,
Vpclmulqdq,
Vtx,
Waitpkg,
Wbnoinvd,
X87,
Xop,
Xsave,
Xsavec,
Xsaves,
}
#[doc(hidden)]
impl TryFrom<ffi::ND_INS_SET> for IsaSet {
type Error = decode_error::DecodeError;
fn try_from(value: ffi::ND_INS_SET) -> Result<Self, Self::Error> {
match value {
ffi::_ND_INS_SET::ND_SET_INVALID => {
Err(decode_error::DecodeError::UnknownInstruction(value as u32))
}
ffi::_ND_INS_SET::ND_SET_3DNOW => Ok(IsaSet::I3dnow),
ffi::_ND_INS_SET::ND_SET_ADX => Ok(IsaSet::Adx),
ffi::_ND_INS_SET::ND_SET_AES => Ok(IsaSet::Aes),
ffi::_ND_INS_SET::ND_SET_AMD => Ok(IsaSet::Amd),
ffi::_ND_INS_SET::ND_SET_AMXBF16 => Ok(IsaSet::Amxbf16),
ffi::_ND_INS_SET::ND_SET_AMXINT8 => Ok(IsaSet::Amxint8),
ffi::_ND_INS_SET::ND_SET_AMXTILE => Ok(IsaSet::Amxtile),
ffi::_ND_INS_SET::ND_SET_AVX => Ok(IsaSet::Avx),
ffi::_ND_INS_SET::ND_SET_AVX2 => Ok(IsaSet::Avx2),
ffi::_ND_INS_SET::ND_SET_AVX2GATHER => Ok(IsaSet::Avx2gather),
ffi::_ND_INS_SET::ND_SET_AVX5124FMAPS => Ok(IsaSet::Avx5124fmaps),
ffi::_ND_INS_SET::ND_SET_AVX5124VNNIW => Ok(IsaSet::Avx5124vnniw),
ffi::_ND_INS_SET::ND_SET_AVX512BF16 => Ok(IsaSet::Avx512bf16),
ffi::_ND_INS_SET::ND_SET_AVX512BITALG => Ok(IsaSet::Avx512bitalg),
ffi::_ND_INS_SET::ND_SET_AVX512BW => Ok(IsaSet::Avx512bw),
ffi::_ND_INS_SET::ND_SET_AVX512CD => Ok(IsaSet::Avx512cd),
ffi::_ND_INS_SET::ND_SET_AVX512DQ => Ok(IsaSet::Avx512dq),
ffi::_ND_INS_SET::ND_SET_AVX512ER => Ok(IsaSet::Avx512er),
ffi::_ND_INS_SET::ND_SET_AVX512F => Ok(IsaSet::Avx512f),
ffi::_ND_INS_SET::ND_SET_AVX512FP16 => Ok(IsaSet::Avx512fp16),
ffi::_ND_INS_SET::ND_SET_AVX512IFMA => Ok(IsaSet::Avx512ifma),
ffi::_ND_INS_SET::ND_SET_AVX512PF => Ok(IsaSet::Avx512pf),
ffi::_ND_INS_SET::ND_SET_AVX512VBMI => Ok(IsaSet::Avx512vbmi),
ffi::_ND_INS_SET::ND_SET_AVX512VBMI2 => Ok(IsaSet::Avx512vbmi2),
ffi::_ND_INS_SET::ND_SET_AVX512VNNI => Ok(IsaSet::Avx512vnni),
ffi::_ND_INS_SET::ND_SET_AVX512VP2INTERSECT => Ok(IsaSet::Avx512vp2intersect),
ffi::_ND_INS_SET::ND_SET_AVX512VPOPCNTDQ => Ok(IsaSet::Avx512vpopcntdq),
ffi::_ND_INS_SET::ND_SET_AVXVNNI => Ok(IsaSet::Avxvnni),
ffi::_ND_INS_SET::ND_SET_BMI1 => Ok(IsaSet::Bmi1),
ffi::_ND_INS_SET::ND_SET_BMI2 => Ok(IsaSet::Bmi2),
ffi::_ND_INS_SET::ND_SET_CET_IBT => Ok(IsaSet::CetIbt),
ffi::_ND_INS_SET::ND_SET_CET_SS => Ok(IsaSet::CetSs),
ffi::_ND_INS_SET::ND_SET_CLDEMOTE => Ok(IsaSet::Cldemote),
ffi::_ND_INS_SET::ND_SET_CLFSH => Ok(IsaSet::Clfsh),
ffi::_ND_INS_SET::ND_SET_CLFSHOPT => Ok(IsaSet::Clfshopt),
ffi::_ND_INS_SET::ND_SET_CLWB => Ok(IsaSet::Clwb),
ffi::_ND_INS_SET::ND_SET_CLZERO => Ok(IsaSet::Clzero),
ffi::_ND_INS_SET::ND_SET_CMPXCHG16B => Ok(IsaSet::Cmpxchg16b),
ffi::_ND_INS_SET::ND_SET_CYRIX => Ok(IsaSet::Cyrix),
ffi::_ND_INS_SET::ND_SET_CYRIX_SMM => Ok(IsaSet::CyrixSmm),
ffi::_ND_INS_SET::ND_SET_ENQCMD => Ok(IsaSet::Enqcmd),
ffi::_ND_INS_SET::ND_SET_F16C => Ok(IsaSet::F16c),
ffi::_ND_INS_SET::ND_SET_FMA => Ok(IsaSet::Fma),
ffi::_ND_INS_SET::ND_SET_FMA4 => Ok(IsaSet::Fma4),
ffi::_ND_INS_SET::ND_SET_FRED => Ok(IsaSet::Fred),
ffi::_ND_INS_SET::ND_SET_FXSAVE => Ok(IsaSet::Fxsave),
ffi::_ND_INS_SET::ND_SET_GFNI => Ok(IsaSet::Gfni),
ffi::_ND_INS_SET::ND_SET_HRESET => Ok(IsaSet::Hreset),
ffi::_ND_INS_SET::ND_SET_I186 => Ok(IsaSet::I186),
ffi::_ND_INS_SET::ND_SET_I286PROT => Ok(IsaSet::I286prot),
ffi::_ND_INS_SET::ND_SET_I286REAL => Ok(IsaSet::I286real),
ffi::_ND_INS_SET::ND_SET_I386 => Ok(IsaSet::I386),
ffi::_ND_INS_SET::ND_SET_I486 => Ok(IsaSet::I486),
ffi::_ND_INS_SET::ND_SET_I486REAL => Ok(IsaSet::I486real),
ffi::_ND_INS_SET::ND_SET_I64 => Ok(IsaSet::I64),
ffi::_ND_INS_SET::ND_SET_I86 => Ok(IsaSet::I86),
ffi::_ND_INS_SET::ND_SET_INVLPGB => Ok(IsaSet::Invlpgb),
ffi::_ND_INS_SET::ND_SET_INVPCID => Ok(IsaSet::Invpcid),
ffi::_ND_INS_SET::ND_SET_KL => Ok(IsaSet::Kl),
ffi::_ND_INS_SET::ND_SET_LKGS => Ok(IsaSet::Lkgs),
ffi::_ND_INS_SET::ND_SET_LONGMODE => Ok(IsaSet::Longmode),
ffi::_ND_INS_SET::ND_SET_LWP => Ok(IsaSet::Lwp),
ffi::_ND_INS_SET::ND_SET_LZCNT => Ok(IsaSet::Lzcnt),
ffi::_ND_INS_SET::ND_SET_MCOMMIT => Ok(IsaSet::Mcommit),
ffi::_ND_INS_SET::ND_SET_MMX => Ok(IsaSet::Mmx),
ffi::_ND_INS_SET::ND_SET_MOVBE => Ok(IsaSet::Movbe),
ffi::_ND_INS_SET::ND_SET_MOVDIR64B => Ok(IsaSet::Movdir64b),
ffi::_ND_INS_SET::ND_SET_MOVDIRI => Ok(IsaSet::Movdiri),
ffi::_ND_INS_SET::ND_SET_MPX => Ok(IsaSet::Mpx),
ffi::_ND_INS_SET::ND_SET_MWAITT => Ok(IsaSet::Mwaitt),
ffi::_ND_INS_SET::ND_SET_PAUSE => Ok(IsaSet::Pause),
ffi::_ND_INS_SET::ND_SET_PCLMULQDQ => Ok(IsaSet::Pclmulqdq),
ffi::_ND_INS_SET::ND_SET_PCONFIG => Ok(IsaSet::Pconfig),
ffi::_ND_INS_SET::ND_SET_PENTIUMREAL => Ok(IsaSet::Pentiumreal),
ffi::_ND_INS_SET::ND_SET_PKU => Ok(IsaSet::Pku),
ffi::_ND_INS_SET::ND_SET_POPCNT => Ok(IsaSet::Popcnt),
ffi::_ND_INS_SET::ND_SET_PPRO => Ok(IsaSet::Ppro),
ffi::_ND_INS_SET::ND_SET_PREFETCH_NOP => Ok(IsaSet::PrefetchNop),
ffi::_ND_INS_SET::ND_SET_PTWRITE => Ok(IsaSet::Ptwrite),
ffi::_ND_INS_SET::ND_SET_RDPID => Ok(IsaSet::Rdpid),
ffi::_ND_INS_SET::ND_SET_RDPMC => Ok(IsaSet::Rdpmc),
ffi::_ND_INS_SET::ND_SET_RDPRU => Ok(IsaSet::Rdpru),
ffi::_ND_INS_SET::ND_SET_RDRAND => Ok(IsaSet::Rdrand),
ffi::_ND_INS_SET::ND_SET_RDSEED => Ok(IsaSet::Rdseed),
ffi::_ND_INS_SET::ND_SET_RDTSCP => Ok(IsaSet::Rdtscp),
ffi::_ND_INS_SET::ND_SET_RDWRFSGS => Ok(IsaSet::Rdwrfsgs),
ffi::_ND_INS_SET::ND_SET_SERIALIZE => Ok(IsaSet::Serialize),
ffi::_ND_INS_SET::ND_SET_SGX => Ok(IsaSet::Sgx),
ffi::_ND_INS_SET::ND_SET_SHA => Ok(IsaSet::Sha),
ffi::_ND_INS_SET::ND_SET_SMAP => Ok(IsaSet::Smap),
ffi::_ND_INS_SET::ND_SET_SMX => Ok(IsaSet::Smx),
ffi::_ND_INS_SET::ND_SET_SNP => Ok(IsaSet::Snp),
ffi::_ND_INS_SET::ND_SET_SSE => Ok(IsaSet::Sse),
ffi::_ND_INS_SET::ND_SET_SSE2 => Ok(IsaSet::Sse2),
ffi::_ND_INS_SET::ND_SET_SSE3 => Ok(IsaSet::Sse3),
ffi::_ND_INS_SET::ND_SET_SSE4 => Ok(IsaSet::Sse4),
ffi::_ND_INS_SET::ND_SET_SSE42 => Ok(IsaSet::Sse42),
ffi::_ND_INS_SET::ND_SET_SSE4A => Ok(IsaSet::Sse4a),
ffi::_ND_INS_SET::ND_SET_SSSE3 => Ok(IsaSet::Ssse3),
ffi::_ND_INS_SET::ND_SET_SVM => Ok(IsaSet::Svm),
ffi::_ND_INS_SET::ND_SET_TBM => Ok(IsaSet::Tbm),
ffi::_ND_INS_SET::ND_SET_TDX => Ok(IsaSet::Tdx),
ffi::_ND_INS_SET::ND_SET_TSX => Ok(IsaSet::Tsx),
ffi::_ND_INS_SET::ND_SET_TSXLDTRK => Ok(IsaSet::Tsxldtrk),
ffi::_ND_INS_SET::ND_SET_UD => Ok(IsaSet::Ud),
ffi::_ND_INS_SET::ND_SET_UINTR => Ok(IsaSet::Uintr),
ffi::_ND_INS_SET::ND_SET_UNKNOWN => Ok(IsaSet::Unknown),
ffi::_ND_INS_SET::ND_SET_VAES => Ok(IsaSet::Vaes),
ffi::_ND_INS_SET::ND_SET_VPCLMULQDQ => Ok(IsaSet::Vpclmulqdq),
ffi::_ND_INS_SET::ND_SET_VTX => Ok(IsaSet::Vtx),
ffi::_ND_INS_SET::ND_SET_WAITPKG => Ok(IsaSet::Waitpkg),
ffi::_ND_INS_SET::ND_SET_WBNOINVD => Ok(IsaSet::Wbnoinvd),
ffi::_ND_INS_SET::ND_SET_X87 => Ok(IsaSet::X87),
ffi::_ND_INS_SET::ND_SET_XOP => Ok(IsaSet::Xop),
ffi::_ND_INS_SET::ND_SET_XSAVE => Ok(IsaSet::Xsave),
ffi::_ND_INS_SET::ND_SET_XSAVEC => Ok(IsaSet::Xsavec),
ffi::_ND_INS_SET::ND_SET_XSAVES => Ok(IsaSet::Xsaves),
}
}
}

@ -0,0 +1,191 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! bddisasm x86/x64 instruction decoder
//!
//! Rust bindings for the [bddisasm](https://github.com/bitdefender/bddisasm) x86/x64 decoder library, built on top
//! of [bddisasm-sys](https://crates.io/crates/bddisasm-sys).
//!
//! It supports all existing x86 instruction, offering a wide range of information about each one, including:
//!
//! - operands (implicit and explicit)
//! - access mode for each operand
//! - CPUID feature flags
//! - CPU modes in which an instruction is valid
//!
//! # Usage
//!
//! Add `bddisasm` to your `Cargo.toml`:
//!
//! ```toml
//! [dependencies]
//! bddisasm = "0.1.0"
//! ```
//!
//! # Examples
//!
//! ## Decoding one instruction
//!
//! Use [DecodedInstruction::decode](crate::decoded_instruction::DecodedInstruction::decode) to decode an instruction
//! from a chunk of code.
//!
//! ```
//! use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode, Mnemonic};
//!
//! let code = vec![0x31, 0xc0];
//! match DecodedInstruction::decode(&code, DecodeMode::Bits32) {
//! Ok(ins) => {
//! assert_eq!(ins.mnemonic(), Mnemonic::Xor);
//! println!("{}", ins);
//! },
//! Err(err) => println!("Unable to decode: {}", err),
//! }
//! ```
//!
//! ## Decoding multiple instructions
//!
//! Use [Decoder](crate::decoder::Decoder) to decode multiple instructions from a chunk of code.
//!
//! ```
//! use bddisasm::decoder::{Decoder, DecodeMode};
//!
//! let code = [
//! // ENCLS
//! 0x0f, 0x01, 0xcf,
//! // MOV rax, qword ptr [rbx+rcx*4+0x1234]
//! 0x48, 0x8b, 0x84, 0x8b, 0x34, 0x12, 0x00, 0x00,
//! // Not a valid instruction
//! 0x0f,
//! // WRMSR
//! 0x0f, 0x30,
//! ];
//! let decoder = Decoder::new(&code, DecodeMode::Bits64, 0x1234);
//!
//! for ins in decoder {
//! match ins {
//! Ok(ins) => println!("{}", ins),
//! Err(e) => println!("{}", e),
//! }
//! }
//! ```
//!
//! This will print:
//!
//! ```text
//! ENCLS
//! MOV rax, qword ptr [rbx+rcx*4+0x1234]
//! the provided input buffer is too small
//! WRMSR
//! ```
//!
//! Use [Decoder::decode_next_with_info](crate::decoder::Decoder::decode_next_with_info) to get information about the
//! offset inside the code chunk at which an instruction was decoded from.
//!
//! ```
//! use bddisasm::decoder::{Decoder, DecodeMode};
//!
//! let code = [
//! // ENCLS
//! 0x0f, 0x01, 0xcf,
//! // MOV rax, qword ptr [rbx+rcx*4+0x1234]
//! 0x48, 0x8b, 0x84, 0x8b, 0x34, 0x12, 0x00, 0x00,
//! // Not a valid instruction
//! 0x0f,
//! // WRMSR
//! 0x0f, 0x30,
//! ];
//! let mut decoder = Decoder::new(&code, DecodeMode::Bits64, 0x1234);
//!
//!
//! // Keep decoding until there's nothing left to decode
//! while let Some((result, offset, _)) = decoder.decode_next_with_info() {
//! match result {
//! Ok(ins) => println!("{:#x} {}", offset, ins),
//! Err(e) => println!("Error: `{}` at offset {:#x}", e, offset),
//! }
//! }
//! ```
//!
//! This will print:
//!
//! ```text
//! 0x0 ENCLS
//! 0x3 MOV rax, qword ptr [rbx+rcx*4+0x1234]
//! Error `the provided input buffer is too small` at offset 0xb
//! 0xc WRMSR
//! ```
//!
//! ## Working with instruction operands
//!
//! Instruction operands can be analyzed using the [operand](crate::operand) module. Rich informaion is offered for
//! each type of operand. Bellow is a minimal example that looks at a memory operand.
//!
//! ```
//! # use bddisasm::decode_error::DecodeError;
//! # fn test() -> Result<(), DecodeError> {
//! use bddisasm::decoded_instruction::{DecodedInstruction, DecodeMode};
//! use bddisasm::operand::OpInfo;
//!
//! // ` MOV rax, qword ptr [rcx+r15*2]`
//! let code = b"\x4a\x8b\x04\x79";
//! let ins = DecodedInstruction::decode(code, DecodeMode::Bits64).unwrap();
//! // Get the operands
//! let operands = ins.operands();
//! // Get the second operand which is the source (`[rcx+r15*2]`)
//! let src = operands[1];
//!
//! println!("Source operand type: {}", src.info);
//! match src.info {
//! OpInfo::Mem(mem) => {
//! if let Some(base) = mem.base {
//! println!("Base register: {}", base);
//! } else {
//! println!("No base register");
//! }
//!
//! if let Some(index) = mem.index {
//! println!("Index register: {}", index);
//! } else {
//! println!("No index register");
//! }
//!
//! if let Some(scale) = mem.scale {
//! println!("Scale: {}", scale);
//! } else {
//! println!("No scale");
//! }
//!
//! if let Some(displacement) = mem.disp {
//! println!("Displacement: {}", displacement);
//! } else {
//! println!("No displacement");
//! }
//! },
//! _ => unreachable!(),
//! }
//! # Ok(())
//! # }
//! ```
//!
//! Will print:
//!
//! ```text
//! Source operand type: memory
//! Base register: 1
//! Index register: 15
//! Scale: 2
//! No displacement
//! ```
pub mod cpu_modes;
pub mod cpuid;
pub mod decode_error;
pub mod decoded_instruction;
pub mod decoder;
pub mod fpu_flags;
pub mod instruction_category;
pub mod isa_set;
pub mod mnemonic;
pub mod operand;
pub mod rflags;
pub mod tuple;

File diff suppressed because it is too large Load Diff

@ -0,0 +1,767 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Operand types and details.
extern crate bddisasm_sys as ffi;
use std::fmt;
/// Describes an address operand.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct OpAddr {
/// The ID of the base segment selector.
///
/// `ES` = 0, `CS` = 1, `SS` = 2, `DS` = 3, `FS` = 4, `GS` = 5.
pub base_seg: u16,
/// Offset inside the segment.
pub offset: u64,
}
#[doc(hidden)]
impl From<ffi::ND_OPDESC_ADDRESS> for OpAddr {
fn from(op: ffi::ND_OPDESC_ADDRESS) -> OpAddr {
OpAddr {
base_seg: op.BaseSeg,
offset: op.Offset,
}
}
}
#[doc(hidden)]
impl OpAddr {
pub(crate) fn new(base_seg: u16, offset: u64) -> Self {
Self { base_seg, offset }
}
}
/// The type of a register.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum OpRegType {
/// The register is a 8/16/32/64 bit general purpose register.
Gpr,
/// The register is a segment register.
Seg,
/// The register is a 80-bit FPU register.
Fpu,
/// The register is a 64-bit MMX register.
Mmx,
/// The register is a 128/256/512 bit SSE vector register.
Sse,
/// The register is a control register.
Cr,
/// The register is a debug register.
Dr,
/// The register is a test register.
Tr,
/// The register is a bound register.
Bnd,
/// The register is a mask register.
Msk,
/// The register is a tile register.
Tile,
/// The register is a model specific register.
Msr,
/// The register is a extended control register.
Xcr,
/// The register is a system register.
Sys,
/// The register is a x87 status/control register.
X87,
/// The register is the MXCSR register.
Mxcsr,
/// The register is the PKRU register.
Pkru,
/// The register is the SSP (Shadow Stack Pointer) register.
Ssp,
/// The register is the FLAGS register.
Flg,
/// The register is the instruction pointer register.
Rip,
/// The register is the User Interrupt Flag.
Uif,
}
#[doc(hidden)]
impl From<ffi::ND_OPDESC_REGISTER> for OpRegType {
fn from(op: ffi::ND_OPDESC_REGISTER) -> OpRegType {
match op.Type {
ffi::_ND_REG_TYPE::ND_REG_NOT_PRESENT => panic!("Unexpected ND_REG_NOT_PRESENT"),
ffi::_ND_REG_TYPE::ND_REG_GPR => OpRegType::Gpr,
ffi::_ND_REG_TYPE::ND_REG_SEG => OpRegType::Seg,
ffi::_ND_REG_TYPE::ND_REG_FPU => OpRegType::Fpu,
ffi::_ND_REG_TYPE::ND_REG_MMX => OpRegType::Mmx,
ffi::_ND_REG_TYPE::ND_REG_SSE => OpRegType::Sse,
ffi::_ND_REG_TYPE::ND_REG_CR => OpRegType::Cr,
ffi::_ND_REG_TYPE::ND_REG_DR => OpRegType::Dr,
ffi::_ND_REG_TYPE::ND_REG_TR => OpRegType::Tr,
ffi::_ND_REG_TYPE::ND_REG_BND => OpRegType::Bnd,
ffi::_ND_REG_TYPE::ND_REG_MSK => OpRegType::Msk,
ffi::_ND_REG_TYPE::ND_REG_TILE => OpRegType::Tile,
ffi::_ND_REG_TYPE::ND_REG_MSR => OpRegType::Msr,
ffi::_ND_REG_TYPE::ND_REG_XCR => OpRegType::Xcr,
ffi::_ND_REG_TYPE::ND_REG_SYS => OpRegType::Sys,
ffi::_ND_REG_TYPE::ND_REG_X87 => OpRegType::X87,
ffi::_ND_REG_TYPE::ND_REG_MXCSR => OpRegType::Mxcsr,
ffi::_ND_REG_TYPE::ND_REG_PKRU => OpRegType::Pkru,
ffi::_ND_REG_TYPE::ND_REG_SSP => OpRegType::Ssp,
ffi::_ND_REG_TYPE::ND_REG_FLG => OpRegType::Flg,
ffi::_ND_REG_TYPE::ND_REG_RIP => OpRegType::Rip,
ffi::_ND_REG_TYPE::ND_REG_UIF => OpRegType::Uif,
}
}
}
impl fmt::Display for OpRegType {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self {
Self::Gpr => write!(f, "GPR"),
Self::Seg => write!(f, "Segment"),
Self::Fpu => write!(f, "FPU"),
Self::Mmx => write!(f, "MMX"),
Self::Sse => write!(f, "SSE"),
Self::Cr => write!(f, "CR"),
Self::Dr => write!(f, "DR"),
Self::Tr => write!(f, "TR"),
Self::Bnd => write!(f, "BND"),
Self::Msk => write!(f, "MSK"),
Self::Tile => write!(f, "Tile"),
Self::Msr => write!(f, "MSR"),
Self::Xcr => write!(f, "XCR"),
Self::Sys => write!(f, "Sys"),
Self::X87 => write!(f, "X87"),
Self::Mxcsr => write!(f, "MXCSR"),
Self::Pkru => write!(f, "PKRU"),
Self::Ssp => write!(f, "SSP"),
Self::Flg => write!(f, "Flg"),
Self::Rip => write!(f, "RIP"),
Self::Uif => write!(f, "UIF"),
}
}
}
/// Describes a register operand.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct OpReg {
/// The type of the register.
pub kind: OpRegType,
/// Indicates the register size, in bytes.
///
/// This may not be equal to the [`Operand::size`] field, as a smaller amount of data may be processed from a
/// register (especially if we have a SSE register or a mask register).
pub size: u32,
// TODO: create an enum for each register, instead of using an ID.
/// The register index/ID.
///
/// Registers are numbered as they are in the Intel documentation. Examples:
///
/// - `AL` = `AX` = `EAX` = `RAX` = 0
/// - `R15B` = `R15W` = `R15D` = `R15` = 15
/// - `XMM0` = 0
/// - `ZMM31` = 31
/// - `ES` = 0
/// - `CS` = 1
///
/// # Remarks
///
/// If [kind](OpReg::kind) is [OpRegType::Gpr](OpRegType::Gpr), the high and low part of 16-bit registers will have
/// the same index (for example, `AH` and `AL`). To differentiate between them use [is_high8](OpReg::is_high8).
pub index: usize,
/// The number of registers accessed, starting with [id](OpReg::index).
pub count: u32,
/// `true` if this is the high part of a 16-bit GPR: `AH`, `CH`, `DH`, or `BH`.
pub is_high8: bool,
/// `true` if this is a block register addressing.
pub is_block: bool,
}
#[doc(hidden)]
impl From<ffi::ND_OPDESC_REGISTER> for OpReg {
fn from(op: ffi::ND_OPDESC_REGISTER) -> OpReg {
let kind = OpRegType::from(op);
let is_high8 = op.IsHigh8();
let index = match kind {
OpRegType::Gpr => {
if is_high8 {
// See `ShemuGetGprValue` in `bdshemu.c`.
op.Reg - 4
} else {
op.Reg
}
}
_ => op.Reg,
} as usize;
OpReg {
kind,
size: op.Size,
index,
count: op.Count,
is_high8,
is_block: op.IsBlock(),
}
}
}
/// Holds extra information for instruction that use VSIB addressing.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Vsib {
/// VSIB element size.
pub vsib_element_size: u8,
/// Number of elements scattered/gathered/prefetched.
pub vsib_element_count: u8,
}
/// Shadow stack access types.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum ShadowStackAccess {
None,
/// Explicit memory operand accessed as shadow stack.
Explicit,
/// Shadow Stack Pointer (`SSP`) used as base for addressing using conventional load/store.
SspLdSt,
/// Shadow Stack Pointer (`SSP`) used as base for addressing using `push`/`pop`.
SspPushPop,
/// Privilege 0 `SSP` (`IA32_PL0_SSP`) used (`SETSSBSY`).
Pl0Ssp,
}
#[doc(hidden)]
impl From<ffi::ND_SHSTK_ACCESS> for ShadowStackAccess {
fn from(acc: ffi::ND_SHSTK_ACCESS) -> ShadowStackAccess {
match acc {
ffi::_ND_SHSTK_ACCESS::ND_SHSTK_NONE => ShadowStackAccess::None,
ffi::_ND_SHSTK_ACCESS::ND_SHSTK_EXPLICIT => ShadowStackAccess::Explicit,
ffi::_ND_SHSTK_ACCESS::ND_SHSTK_SSP_LD_ST => ShadowStackAccess::SspLdSt,
ffi::_ND_SHSTK_ACCESS::ND_SHSTK_SSP_PUSH_POP => ShadowStackAccess::SspPushPop,
ffi::_ND_SHSTK_ACCESS::ND_SHSTK_PL0_SSP => ShadowStackAccess::Pl0Ssp,
// NOTE: when updating this take care to also update the `From<u8>` implementation!
// TODO: any way of keeping these in sync automagically?
}
}
}
impl From<u8> for ShadowStackAccess {
fn from(acc: u8) -> ShadowStackAccess {
if acc == ffi::_ND_SHSTK_ACCESS::ND_SHSTK_NONE as u8 {
ShadowStackAccess::None
} else if acc == ffi::_ND_SHSTK_ACCESS::ND_SHSTK_EXPLICIT as u8 {
ShadowStackAccess::Explicit
} else if acc == ffi::_ND_SHSTK_ACCESS::ND_SHSTK_SSP_LD_ST as u8 {
ShadowStackAccess::SspLdSt
} else if acc == ffi::_ND_SHSTK_ACCESS::ND_SHSTK_SSP_PUSH_POP as u8 {
ShadowStackAccess::SspPushPop
} else if acc == ffi::_ND_SHSTK_ACCESS::ND_SHSTK_PL0_SSP as u8 {
ShadowStackAccess::Pl0Ssp
} else {
panic!("Unexpected shadow stack access type: {}", acc)
}
}
}
/// Describes a memory operand.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct OpMem {
/// `true` if the memory operand is a broadcast operand.
pub has_broadcast: bool,
/// `true` if this is a rip-relative addressing. Base, Index, Scale are all ignored.
pub is_rip_rel: bool,
/// `true` if this is a stack op.
///
/// # Note
///
/// Explicit stack accesses are not included (eg: `mov eax, [rsp]`).
pub is_stack: bool,
/// `true` for `[RSI]` and `[RDI]` operands inside string operations.
pub is_string: bool,
/// `true` if direct addressing (`MOV [...], EAX, 0xA3`).
pub is_direct: bool,
/// `true` if this is a bit base. Used for `BT*` instructions.
///
/// The bitbase stored in the second operand must be added to the linear address.
pub is_bitbase: bool,
/// `true` if the memory operand is address generation and no memory access is made.
pub is_ag: bool,
/// `true` MIB addressing is used (MPX instructions).
pub is_mib: bool,
/// `true` the addressing uses sibmem (AMX instructions).
pub is_sib_mem: bool,
/// Base segment used to address the memory.
pub seg: Option<u8>,
/// Base register.
pub base: Option<u8>,
/// Base register size, in bytes. Max 8 bytes.
pub base_size: Option<u32>,
/// Index register. Can be a vector reg (ZMM0-ZMM31).
pub index: Option<u8>,
/// Index register size, in bytes. Max 8 bytes.
pub index_size: Option<u32>,
/// The `index` register selects a vector register.
pub vsib: Option<Vsib>,
/// Scale: 1, 2, 4 or 8. Always present if `index` is present.
pub scale: Option<u8>,
/// Sign extended displacement.
pub disp: Option<u64>,
/// Displacement size. Max 4 bytes.
pub disp_size: Option<u8>,
/// Compressed displacement size - 1, 2, 4, 8, 16, 32, 64.
pub comp_disp_size: Option<u8>,
/// Shadow stack access type.
///
/// Will be `None` if the shadow stack is not accessed.
pub shadow_stack_access: Option<ShadowStackAccess>,
}
#[doc(hidden)]
impl From<ffi::ND_OPDESC_MEMORY> for OpMem {
fn from(op: ffi::ND_OPDESC_MEMORY) -> OpMem {
let seg = if op.HasSeg() { Some(op.Seg) } else { None };
let (base, base_size) = if op.HasBase() {
(Some(op.Base), Some(op.BaseSize))
} else {
(None, None)
};
let (index, index_size, scale) = if op.HasIndex() {
(Some(op.Index), Some(op.IndexSize), Some(op.Scale))
} else {
(None, None, None)
};
let (disp, disp_size) = if op.HasDisp() {
(Some(op.Disp), Some(op.DispSize))
} else {
(None, None)
};
let comp_disp_size = if op.HasCompDisp() {
Some(op.CompDispSize)
} else {
None
};
let (vsib, index_size) = if op.IsVsib() {
(
Some(Vsib {
vsib_element_size: op.Vsib.ElemSize,
vsib_element_count: op.Vsib.ElemCount,
}),
Some(op.Vsib.IndexSize as u32),
)
} else {
(None, index_size)
};
let shadow_stack_access = if op.IsShadowStack() {
Some(ShadowStackAccess::from(op.ShStkType))
} else {
None
};
OpMem {
has_broadcast: op.HasBroadcast(),
is_rip_rel: op.IsRipRel(),
is_stack: op.IsStack(),
is_string: op.IsString(),
is_direct: op.IsDirect(),
is_bitbase: op.IsBitbase(),
is_ag: op.IsAG(),
is_mib: op.IsMib(),
is_sib_mem: op.IsSibMem(),
seg,
base,
base_size,
index,
index_size,
vsib,
scale,
disp,
disp_size,
comp_disp_size,
shadow_stack_access,
}
}
}
/// Extended operand information, based on the actual type of the operand.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum OpInfo {
None,
/// The operand is a register.
Reg(OpReg),
/// The operand is located in memory.
Mem(OpMem),
/// The operand is an immediate.
Imm(u64),
/// The operand is a relative offset.
Offs(u64),
/// The operand is an absolute address, in the form `seg:offset`.
Addr(OpAddr),
/// The operand is an implicit constant.
Const(u64),
/// An entire bank/set of registers are being accessed. Used in `PUSHA`/`POPA`/`XSAVE`/`LOADALL`.
Bank,
}
#[doc(hidden)]
impl From<ffi::ND_OPERAND> for OpInfo {
fn from(op: ffi::ND_OPERAND) -> OpInfo {
match op.Type {
ffi::_ND_OPERAND_TYPE::ND_OP_NOT_PRESENT => OpInfo::None,
ffi::_ND_OPERAND_TYPE::ND_OP_REG => {
OpInfo::Reg(OpReg::from(unsafe { op.Info.Register }))
}
ffi::_ND_OPERAND_TYPE::ND_OP_MEM => OpInfo::Mem(OpMem::from(unsafe { op.Info.Memory })),
ffi::_ND_OPERAND_TYPE::ND_OP_IMM => OpInfo::Imm(unsafe { op.Info.Immediate }.Imm),
ffi::_ND_OPERAND_TYPE::ND_OP_OFFS => {
OpInfo::Offs(unsafe { op.Info.RelativeOffset }.Rel)
}
ffi::_ND_OPERAND_TYPE::ND_OP_ADDR => {
OpInfo::Addr(OpAddr::from(unsafe { op.Info.Address }))
}
ffi::_ND_OPERAND_TYPE::ND_OP_CONST => OpInfo::Const(unsafe { op.Info.Constant }.Const),
ffi::_ND_OPERAND_TYPE::ND_OP_BANK => OpInfo::Bank,
}
}
}
impl fmt::Display for OpInfo {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self {
Self::None => write!(f, "None"),
Self::Reg(_) => write!(f, "register"),
Self::Mem(_) => write!(f, "memory"),
Self::Imm(_) => write!(f, "immediate"),
Self::Offs(_) => write!(f, "relative offset"),
Self::Addr(_) => write!(f, "absolute address"),
Self::Const(_) => write!(f, "implicit constant"),
Self::Bank => write!(f, "register bank"),
}
}
}
/// Operand size.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum OpSize {
/// The inner value contains the operand size in bytes.
Bytes(usize),
/// The size of a cache line.
CacheLine,
/// Unknown size.
Unknown,
}
impl fmt::Display for OpSize {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self {
Self::Bytes(size) => write!(f, "{} bytes", size),
Self::CacheLine => write!(f, "cache line"),
Self::Unknown => write!(f, "unknown"),
}
}
}
#[doc(hidden)]
impl From<ffi::ND_OPERAND_SIZE> for OpSize {
fn from(sz: ffi::ND_OPERAND_SIZE) -> OpSize {
match sz {
ffi::ND_SIZE_8BIT => OpSize::Bytes(1),
ffi::ND_SIZE_16BIT => OpSize::Bytes(2),
ffi::ND_SIZE_32BIT => OpSize::Bytes(4),
ffi::ND_SIZE_48BIT => OpSize::Bytes(6),
ffi::ND_SIZE_64BIT => OpSize::Bytes(8),
ffi::ND_SIZE_80BIT => OpSize::Bytes(10),
ffi::ND_SIZE_112BIT => OpSize::Bytes(14),
ffi::ND_SIZE_128BIT => OpSize::Bytes(16),
ffi::ND_SIZE_224BIT => OpSize::Bytes(28),
ffi::ND_SIZE_256BIT => OpSize::Bytes(32),
ffi::ND_SIZE_384BIT => OpSize::Bytes(48),
ffi::ND_SIZE_512BIT => OpSize::Bytes(64),
ffi::ND_SIZE_752BIT => OpSize::Bytes(94),
ffi::ND_SIZE_864BIT => OpSize::Bytes(108),
ffi::ND_SIZE_4096BIT => OpSize::Bytes(512),
ffi::ND_SIZE_1KB => OpSize::Bytes(1024),
ffi::ND_SIZE_CACHE_LINE => OpSize::CacheLine,
ffi::ND_SIZE_UNKNOWN => OpSize::Unknown,
_ => panic!("Unespected operand size: {}", sz),
}
}
}
/// Operand access mode.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct OpAccess {
/// The operand is read.
pub read: bool,
/// The operand is written.
pub write: bool,
/// The operand is read only under some conditions.
pub cond_read: bool,
/// The operand is written only under some conditions.
pub cond_write: bool,
/// The operand is prefetched.
pub prefetch: bool,
}
#[doc(hidden)]
impl From<ffi::ND_OPERAND_ACCESS> for OpAccess {
fn from(acc: ffi::ND_OPERAND_ACCESS) -> OpAccess {
let acc = unsafe { acc.__bindgen_anon_1 };
OpAccess {
read: acc.Read() != 0,
write: acc.Write() != 0,
cond_read: acc.CondRead() != 0,
cond_write: acc.CondWrite() != 0,
prefetch: acc.Prefetch() != 0,
}
}
}
/// Broadcast specifier.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Broadcast {
/// Number of times to broadcast the element.
pub count: u8,
/// Size of one element.
pub size: u8,
}
/// Decorator information.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Decorator {
/// If mask is present, holds the ID of the mask register (`K0` - `K7`) used.
pub mask_register: Option<u8>,
/// `true` if zeroing will be made, `false` if merging will be made.
pub has_zero: bool,
/// If broadcasting is being made, holds the broadcast specifier.
pub broadcast: Option<Broadcast>,
}
#[doc(hidden)]
impl From<ffi::ND_OPERAND_DECORATOR> for Decorator {
fn from(decorator: ffi::ND_OPERAND_DECORATOR) -> Decorator {
let mask_register = if decorator.HasMask() {
Some(decorator.Mask.Msk)
} else {
None
};
let broadcast = if decorator.HasBroadcast() {
Some(Broadcast {
count: decorator.Broadcast.Count,
size: decorator.Broadcast.Size,
})
} else {
None
};
Decorator {
mask_register,
has_zero: decorator.HasZero(),
broadcast,
}
}
}
/// Describes an instruction operand.
///
/// Each operand type encodes different information. See [`OpInfo`] for details.
///
/// # Examples
///
/// ```
/// # use std::error::Error;
/// #
/// # fn main() -> Result<(), Box<dyn Error>> {
/// use bddisasm::decoder::{DecodedInstruction, DecodeMode};
/// use bddisasm::operand::*;
///
/// // `MOV ah, byte ptr [rcx+rdx*2+0x8]`
/// let code = vec![0x8a, 0x64, 0x51, 0x08];
/// let ins = DecodedInstruction::decode(&code, DecodeMode::Bits64)?;
///
/// let operands = ins.operands();
/// let dst = operands[0];
/// let src = operands[1];
///
/// // Get the size of each operand
/// println!("Destination size: {}", dst.size);
/// println!("Source size: {}", src.size);
///
/// // Get the type of the destination operand
/// match dst.info {
/// OpInfo::Reg(reg) => {
/// println!("Register kind: {} Size: {} Index: {}", reg.kind, reg.size, reg.index)
/// },
/// // In this case we know that the destination operand is a register
/// _ => println!("Unexpected operand info type"),
/// }
///
/// # Ok(())
/// # }
/// ```
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub struct Operand {
/// Extended operand information.
pub info: OpInfo,
// Operand size in bytes.
//
// This should be used when operating with the operand. It includes sign-extension or zero-extensions.
//
// # Note
///
/// It indicates the actual amount of data used for processing. If the operand is a [register](OpInfo::Reg), it
/// MAY NOT indicate the register size. Use the [info](Operand::info) field to get the actual register size in these
/// cases.
pub size: OpSize,
/// Raw size inside the instruction.
///
/// This will usually be identical to [size](Operand::size), however, some instructions force the actual size of
/// their operands to 64 bit (`PUSH`/`POP` or branches are good examples).
///
/// Although the raw size of the relative offset or the immediate will be [raw_size](Operand::raw_size), internally,
/// the CPU will use [size](Operand::size) (usually sign-extended).
pub raw_size: OpSize,
/// Access mode.
pub access: OpAccess,
/// `true` if the operand is default. This also applies to implicit operands.
pub is_default: bool,
/// Decorator information.
pub decorator: Decorator,
}
#[doc(hidden)]
impl From<ffi::ND_OPERAND> for Operand {
fn from(op: ffi::ND_OPERAND) -> Operand {
Operand {
info: OpInfo::from(op),
size: OpSize::from(op.Size),
raw_size: OpSize::from(op.RawSize),
access: OpAccess::from(op.Access),
is_default: unsafe { op.Flags.__bindgen_anon_1 }.IsDefault() != 0,
decorator: Decorator::from(op.Decorator),
}
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::decoded_instruction::{DecodeMode, DecodedInstruction};
#[test]
fn ops() {
let code = vec![0x8a, 0x64, 0x51, 0x08];
let ins = DecodedInstruction::decode(&code, DecodeMode::Bits64).unwrap();
let operands = ins.operands();
assert_eq!(operands.len(), 2);
let dest = operands[0];
assert_eq!(dest.size, OpSize::Bytes(1));
assert_eq!(dest.raw_size, OpSize::Bytes(1));
assert_eq!(dest.is_default, false);
assert!(dest.access.write);
if let OpInfo::Reg(dst_reg) = dest.info {
assert_eq!(dst_reg.kind, OpRegType::Gpr);
assert_eq!(dst_reg.size, 1);
assert_eq!(dst_reg.index, 0);
assert_eq!(dst_reg.is_high8, true);
} else {
unreachable!();
}
let src = operands[1];
assert_eq!(src.size, OpSize::Bytes(1));
assert_eq!(src.raw_size, OpSize::Bytes(1));
assert_eq!(src.is_default, false);
assert!(src.access.read);
if let OpInfo::Mem(src_mem) = src.info {
assert_eq!(src_mem.seg, Some(3));
assert_eq!(src_mem.base, Some(1));
assert_eq!(src_mem.base_size, Some(8));
assert_eq!(src_mem.index, Some(2));
assert_eq!(src_mem.index_size, Some(8));
assert_eq!(src_mem.scale, Some(2));
assert_eq!(src_mem.disp, Some(8));
assert_eq!(src_mem.disp_size, Some(1));
} else {
unreachable!();
}
}
}

@ -0,0 +1,46 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Describes the bits in the FLAGS register.
extern crate bddisasm_sys as ffi;
/// Carry flag.
pub const CF: u32 = ffi::NDR_RFLAG_CF;
/// Parity flag.
pub const PF: u32 = ffi::NDR_RFLAG_PF;
/// Auxiliary flag.
pub const AF: u32 = ffi::NDR_RFLAG_AF;
/// Zero flag.
pub const ZF: u32 = ffi::NDR_RFLAG_ZF;
/// Sign flag.
pub const SF: u32 = ffi::NDR_RFLAG_SF;
/// Trap flag.
pub const TF: u32 = ffi::NDR_RFLAG_TF;
/// Interrupt flag.
pub const IF: u32 = ffi::NDR_RFLAG_IF;
/// Direction flag.
pub const DF: u32 = ffi::NDR_RFLAG_DF;
/// Overflow flag.
pub const OF: u32 = ffi::NDR_RFLAG_OF;
/// I/O privilege level flag.
pub const IOPL: u32 = ffi::NDR_RFLAG_IOPL;
/// Nested task flag.
pub const NT: u32 = ffi::NDR_RFLAG_NT;
/// Resume flag.
pub const RF: u32 = ffi::NDR_RFLAG_RF;
/// Virtual mode flag.
pub const VM: u32 = ffi::NDR_RFLAG_VM;
/// Alignment check flag.
pub const AC: u32 = ffi::NDR_RFLAG_AC;
/// Virtual interrupts flag.
pub const VIF: u32 = ffi::NDR_RFLAG_VIF;
/// Virtual interrupt pending flag.
pub const VIP: u32 = ffi::NDR_RFLAG_VIP;
/// CPUID identification flag.
pub const ID: u32 = ffi::NDR_RFLAG_ID;
pub(crate) fn flags_raw(flags: ffi::ND_RFLAGS) -> u32 {
unsafe { flags.Raw }
}

@ -0,0 +1,122 @@
/*
* Copyright (c) 2021 Bitdefender
* SPDX-License-Identifier: Apache-2.0
*/
//! Instruction tuple type.
extern crate bddisasm_sys as ffi;
/// Instruction tuple type.
///
/// Used to determine compressed displacement size for `disp8 EVEX` instructions. Note that most of the `EVEX` encoded
/// instructions use the compressed displacement addressing scheme.
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
pub enum Tuple {
None,
/// Full Vector.
Fv,
/// Half Vector.
Hv,
/// Quarter Vector.
Qv,
/// Tuple1 scalar, size 8 bit.
T1S8,
/// Tuple1 scalar, size 16 bit.
T1S16,
/// Tuple1 scalar, size 32/64 bit.
T1S,
/// Tuple1 float, size 32/64 bit.
T1F,
/// Tuple2, 64/128 bit.
T2,
/// Tuple4, 128/256 bit.
T4,
/// Tuple8, 256 bit.
T8,
/// Full Vector Memory.
Fvm,
/// Half Vector Memory.
Hvm,
/// Quarter Vector Memory.
Qvm,
/// Oct Vector Memory.
OVm,
/// M128, 128 bit.
M128,
/// DUP (VMOVDDUP).
Dup,
/// 4 x 32 bit Memory Elements are referenced.
T14X,
}
#[doc(hidden)]
impl From<ffi::ND_TUPLE> for Tuple {
fn from(value: ffi::ND_TUPLE) -> Tuple {
match value {
ffi::_ND_TUPLE::ND_TUPLE_None => Tuple::None,
ffi::_ND_TUPLE::ND_TUPLE_FV => Tuple::Fv,
ffi::_ND_TUPLE::ND_TUPLE_HV => Tuple::Hv,
ffi::_ND_TUPLE::ND_TUPLE_QV => Tuple::Qv,
ffi::_ND_TUPLE::ND_TUPLE_T1S8 => Tuple::T1S8,
ffi::_ND_TUPLE::ND_TUPLE_T1S16 => Tuple::T1S16,
ffi::_ND_TUPLE::ND_TUPLE_T1S => Tuple::T1S,
ffi::_ND_TUPLE::ND_TUPLE_T1F => Tuple::T1F,
ffi::_ND_TUPLE::ND_TUPLE_T2 => Tuple::T2,
ffi::_ND_TUPLE::ND_TUPLE_T4 => Tuple::T4,
ffi::_ND_TUPLE::ND_TUPLE_T8 => Tuple::T8,
ffi::_ND_TUPLE::ND_TUPLE_FVM => Tuple::Fvm,
ffi::_ND_TUPLE::ND_TUPLE_HVM => Tuple::Hvm,
ffi::_ND_TUPLE::ND_TUPLE_QVM => Tuple::Qvm,
ffi::_ND_TUPLE::ND_TUPLE_OVM => Tuple::OVm,
ffi::_ND_TUPLE::ND_TUPLE_M128 => Tuple::M128,
ffi::_ND_TUPLE::ND_TUPLE_DUP => Tuple::Dup,
ffi::_ND_TUPLE::ND_TUPLE_T1_4X => Tuple::T14X,
// NOTE: when updating this take care to also update the `From<u32>` implementation!
// TODO: any way of keeping these in sync automagically?
}
}
}
impl From<u32> for Tuple {
fn from(value: u32) -> Tuple {
if value == ffi::_ND_TUPLE::ND_TUPLE_None as u32 {
Tuple::None
} else if value == ffi::_ND_TUPLE::ND_TUPLE_FV as u32 {
Tuple::Fv
} else if value == ffi::_ND_TUPLE::ND_TUPLE_HV as u32 {
Tuple::Hv
} else if value == ffi::_ND_TUPLE::ND_TUPLE_QV as u32 {
Tuple::Qv
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T1S8 as u32 {
Tuple::T1S8
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T1S16 as u32 {
Tuple::T1S16
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T1S as u32 {
Tuple::T1S
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T1F as u32 {
Tuple::T1F
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T2 as u32 {
Tuple::T2
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T4 as u32 {
Tuple::T4
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T8 as u32 {
Tuple::T8
} else if value == ffi::_ND_TUPLE::ND_TUPLE_FVM as u32 {
Tuple::Fvm
} else if value == ffi::_ND_TUPLE::ND_TUPLE_HVM as u32 {
Tuple::Hvm
} else if value == ffi::_ND_TUPLE::ND_TUPLE_QVM as u32 {
Tuple::Qvm
} else if value == ffi::_ND_TUPLE::ND_TUPLE_OVM as u32 {
Tuple::OVm
} else if value == ffi::_ND_TUPLE::ND_TUPLE_M128 as u32 {
Tuple::M128
} else if value == ffi::_ND_TUPLE::ND_TUPLE_DUP as u32 {
Tuple::Dup
} else if value == ffi::_ND_TUPLE::ND_TUPLE_T1_4X as u32 {
Tuple::T14X
} else {
panic!("Unknown tuple: {}", value)
}
}
}
Loading…
Cancel
Save