Add some comments

This commit is contained in:
Quentin Legot 2023-04-03 15:26:55 +02:00
parent 8c844c3e5c
commit 025ede6080

View File

@ -1,5 +1,4 @@
use crate::Machine; use crate::Machine;
use std::fmt::Error;
use std::fs; use std::fs;
use std::io::Read; use std::io::Read;
@ -24,65 +23,117 @@ pub fn load(path: &str, machine: &mut Machine, start_index: usize) -> Result<(),
instructions.push(u32::from_le_bytes(buf)); instructions.push(u32::from_le_bytes(buf));
} }
} }
for i in 0..instructions.len() { for (i, inst) in instructions.iter().enumerate() {
machine.write_memory(4, 4 * i + start_index, instructions[i] as u64); machine.write_memory(4, 4 * i + start_index, inst.to_owned() as u64);
} }
// #[cfg(debug_assertions)] // #[cfg(debug_assertions)]
// println!("{:04x?}", instructions); // only print loaded program in debug build // println!("{:04x?}", instructions); // only print loaded program in debug build
Ok(()) Ok(())
} }
/// The elf header defines principes aspects of the binary files, it's place at the start of the file
/// see <https://en.wikipedia.org/wiki/Executable_and_Linkable_Format#File_header> for more informations
pub struct ElfHeader { pub struct ElfHeader {
/// Defines whether the file is big or little endian
/// true correspond to big endian, false otherwise
///
/// Offset: 0x05, size: 1 byte
pub endianess: bool, pub endianess: bool,
/// Defines whether the file is 32 bits or 64 bits
///
/// Offset: 0x04, size: 1 byte
pub is_32bits: bool, pub is_32bits: bool,
/// Version of the elf file, current version is 1
///
/// Offset: 0x06, size: 1 byte
pub version: u8, pub version: u8,
/// Identifies the target ABI.
///
/// In this implementation: Defines if the target abi is system V compliant
///
/// Offset: 0x07, size: 1 byte
pub sys_v_abi: bool, pub sys_v_abi: bool,
/// Identifies target ISA, 0xF3 correspond to RISC-V
///
/// In this implementatio, true if target isa is RISC-V, false otherwise
///
/// Offset: 0x12, size: 2 bytes
pub is_riscv_target: bool, pub is_riscv_target: bool,
/// Memory address of the entry point from w<here the process starts its execution.
/// If the program doesn't have an entrypoint (i.e. not an executable), the value is 0
///
/// Offset: 0x18, size: 4 (32 bits) or 8 (64 bits)
pub entrypoint: u64, pub entrypoint: u64,
/// Size of the elf header, 64 bytes for 64 bits and 52 for 32 bits
///
/// Offset: 0x28(32 bits) or 0x34 (64 bits), size: 2 bytes
pub elf_header_size: u16, pub elf_header_size: u16,
/// Position of the first program header entry
///
/// Offset: 0x1C (32 bits) or 0x20 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub program_header_location: u64, pub program_header_location: u64,
/// Number of entries in the progream header table
///
/// Offset: 0x2C (32 bits) or 0x38 (64 bits), size: 2 bytes
pub program_header_entries: u16, pub program_header_entries: u16,
/// Size of a program header entry
///
/// Offset: 0x2A (32 bits) or 0x36 (64 bits), size: 2 bytes
pub program_header_size: u16, pub program_header_size: u16,
/// Position of the first section header entry
///
/// Offset: 0x20 (32 bits) or 0x28 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub section_header_location: u64, pub section_header_location: u64,
/// Number of entries in the section header table
///
/// Offset: 0x30 (32 bits) or 0x3C (64 bits), size: 2 bytes
pub section_header_entries: u16, pub section_header_entries: u16,
/// Size of a section header entry
///
/// Offset: 0x2E (32 bits) or 0x36 (64 bits), size: 2 bytes
pub section_header_size: u16, pub section_header_size: u16,
} }
impl ElfHeader { impl ElfHeader {
/// return true if the 4 first bytes constitude the elf magic number /// return true if the 4 first bytes constitude the elf magic number
fn is_elf(instructions: &Vec<u8>) -> bool { fn is_elf(instructions: &[u8]) -> bool {
instructions.get(0..4) == Option::Some(&[0x7f, 0x45, 0x4c, 0x46]) instructions.get(0..4) == Option::Some(&[0x7f, 0x45, 0x4c, 0x46])
} }
/// return true if big endian, false otherwise /// return true if big endian, false otherwise
fn check_endianess(instructions: &Vec<u8>) -> bool { fn check_endianess(instructions: &[u8]) -> bool {
instructions.get(5) == Option::Some(&2) instructions.get(5) == Option::Some(&2)
} }
/// return true if file is 32 bits, false if 64 bits /// return true if file is 32 bits, false if 64 bits
fn is_32bits(instructions: &Vec<u8>) -> bool { fn is_32bits(instructions: &[u8]) -> bool {
instructions.get(4) == Option::Some(&1) instructions.get(4) == Option::Some(&1)
} }
/// return the version of the elf file (should be 1) /// return the version of the elf file (should be 1)
/// Can be None if the file is smaller than 7 bytes -> the file is invalid /// Can be None if the file is smaller than 7 bytes -> the file is invalid
fn get_version(instructions: &Vec<u8>) -> Option<u8> { fn get_version(instructions: &[u8]) -> Option<u8> {
instructions.get(6).copied() // work as primitives implements Copy instructions.get(6).copied() // work as primitives implements Copy
} }
/// return true if target abi of the binary file is System V, false otherwise /// return true if target abi of the binary file is System V, false otherwise
fn is_system_v_elf(instructions: &Vec<u8>) -> bool { fn is_system_v_elf(instructions: &[u8]) -> bool {
instructions.get(7) == Option::Some(&0) instructions.get(7) == Option::Some(&0)
} }
/// return true if specified target instruction set architecture is RISCV /// return true if specified target instruction set architecture is RISCV
fn is_riscv_isa(instructions: &Vec<u8>) -> bool { fn is_riscv_isa(instructions: &[u8]) -> bool {
Self::get_u16_value(instructions, 0x12) == Option::Some(0xf3) Self::get_u16_value(instructions, 0x12) == Option::Some(0xf3)
} }
/// memory address of the entry point from where the process starts its execution /// memory address of the entry point from where the process starts its execution
fn get_entrypoint(instructions: &Vec<u8>, is_32bits: bool) -> Option<u64> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_entrypoint(instructions: &[u8], is_32bits: bool) -> Option<u64> {
if is_32bits { if is_32bits {
get_address_point(instructions, 0x18, true) get_address_point(instructions, 0x18, true)
} else { } else {
@ -91,7 +142,12 @@ impl ElfHeader {
} }
/// Memory address of the start of the program header table /// Memory address of the start of the program header table
fn get_program_header_table_location(instructions: &Vec<u8>, is_32bits: bool) -> Option<u64> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_program_header_table_location(instructions: &[u8], is_32bits: bool) -> Option<u64> {
if is_32bits { if is_32bits {
get_address_point(instructions, 0x1c, true) get_address_point(instructions, 0x1c, true)
} else { } else {
@ -99,8 +155,13 @@ impl ElfHeader {
} }
} }
/// Memory address of the start of the section header table /// Memory address of the start of the section header table
fn get_section_header_table_location(instructions: &Vec<u8>, is_32bits: bool) -> Option<u64> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_section_header_table_location(instructions: &[u8], is_32bits: bool) -> Option<u64> {
if is_32bits { if is_32bits {
get_address_point(instructions, 0x20, true) get_address_point(instructions, 0x20, true)
} else { } else {
@ -109,37 +170,69 @@ impl ElfHeader {
} }
/// Return the size of the header, normally, 0x40 for 64 bits bin and 0x34 for 32 bits /// Return the size of the header, normally, 0x40 for 64 bits bin and 0x34 for 32 bits
fn get_elf_header_size(instructions: &Vec<u8>, is_32bits: bool) -> Option<u16> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_elf_header_size(instructions: &[u8], is_32bits: bool) -> Option<u16> {
let address = if is_32bits { 0x28 } else { 0x34 }; let address = if is_32bits { 0x28 } else { 0x34 };
Self::get_u16_value(instructions, address) Self::get_u16_value(instructions, address)
} }
/// return the size of a program header table entry /// return the size of a program header table entry
fn get_program_header_size(instructions: &Vec<u8>, is_32bits: bool) -> Option<u16> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_program_header_size(instructions: &[u8], is_32bits: bool) -> Option<u16> {
let address = if is_32bits { 0x2a } else { 0x36 }; let address = if is_32bits { 0x2a } else { 0x36 };
Self::get_u16_value(instructions, address) Self::get_u16_value(instructions, address)
} }
/// return the number of entries in the program header /// return the number of entries in the program header
fn get_number_entries_program_header(instructions: &Vec<u8>, is_32bits: bool) -> Option<u16> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_number_entries_program_header(instructions: &[u8], is_32bits: bool) -> Option<u16> {
let address = if is_32bits { 0x2c } else { 0x38 }; let address = if is_32bits { 0x2c } else { 0x38 };
Self::get_u16_value(instructions, address) Self::get_u16_value(instructions, address)
} }
/// Return the size of a section header table entry /// Return the size of a section header table entry
fn get_section_header_size(instructions: &Vec<u8>, is_32bits: bool) -> Option<u16> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_section_header_size(instructions: &[u8], is_32bits: bool) -> Option<u16> {
let address = if is_32bits { 0x2e } else { 0x3a }; let address = if is_32bits { 0x2e } else { 0x3a };
Self::get_u16_value(instructions, address) Self::get_u16_value(instructions, address)
} }
/// Return the number of entries in the section header /// Return the number of entries in the section header
fn get_section_header_num_entries(instructions: &Vec<u8>, is_32bits: bool) -> Option<u16> { ///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **is_32bits** defines whether the binary file is 32 bits or 64 bits
fn get_section_header_num_entries(instructions: &[u8], is_32bits: bool) -> Option<u16> {
let address = if is_32bits { 0x30 } else { 0x3c }; let address = if is_32bits { 0x30 } else { 0x3c };
Self::get_u16_value(instructions, address) Self::get_u16_value(instructions, address)
} }
/// Return a u16 value, usually for the size or the number of entries inside a header /// Return a u16 value, usually for the size or the number of entries inside a header
fn get_u16_value(instructions: &Vec<u8>, address: usize) -> Option<u16> { ///
/// This method retrieve 2 bytes and concatenate them assuming the file is little endian
///
/// ## Arguments:
///
/// **instructions** List of bytes of the loaded binary file
/// **address** Position of the first byte
fn get_u16_value(instructions: &[u8], address: usize) -> Option<u16> {
let mut bytes: [u8; 2] = [0; 2]; let mut bytes: [u8; 2] = [0; 2];
bytes[0] = instructions.get(address).copied()?; bytes[0] = instructions.get(address).copied()?;
bytes[1] = instructions.get(address + 1).copied()?; bytes[1] = instructions.get(address + 1).copied()?;
@ -188,27 +281,63 @@ impl TryFrom<&Vec<u8>> for ElfHeader {
} }
/// Flag of a section, a section can have multiples flags by adding the values
#[allow(clippy::enum_variant_names)]
#[allow(dead_code)]
pub enum FlagValue { pub enum FlagValue {
/// The section is writable
ShfWrite = 0x1, ShfWrite = 0x1,
/// The section need to be allocate/occupe memory during execution
ShfAlloc = 0x2, ShfAlloc = 0x2,
/// The section need to be executable
ShfExecinstr = 0x4, ShfExecinstr = 0x4,
/// Section might ber merged
ShfMerge = 0x10, ShfMerge = 0x10,
/// Contain null-terminated (\0) strings
ShfStrings = 0x20, ShfStrings = 0x20,
ShfInfoLink = 0x40,
ShfLinkOrder = 0x80,
// There is others but are unrelevant (I think) // There is others but are unrelevant (I think)
} }
/// Section header entry, contains useful informations for each sections of the binary file
///
/// see <https://en.wikipedia.org/wiki/Executable_and_Linkable_Format#Section_header>
#[derive(Debug)] #[derive(Debug)]
pub struct SectionHeader { pub struct SectionHeader {
/// Offset to a string in .shstrtab section that represent the name of this section
///
/// Offset: 0x0, size: 4 bytes
pub name_offset: u32, pub name_offset: u32,
/// Identify the type of this header
///
/// Offset: 0x4, size: 4 bytes
pub header_type: u32, pub header_type: u32,
/// Identify the atributes of this section
///
/// see `Self::does_flag_contains_key(self, FlagValue)`
///
/// Offset: 0x8, size: 4 (32 bits) or 8 (64 bits) bytes
pub flags: u64, pub flags: u64,
/// Virtual address of the section in memory if section is loaded, 0x0 otherwise
///
/// Offset: 0x0C (32 bits) or 0x10 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub virt_addr: u64, pub virt_addr: u64,
/// Offset of the section in the file image (binary file)
///
/// Offset: 0x10 (32 bits) or 0x18 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub image_offset: u64, pub image_offset: u64,
/// Size of the section in the file image, may be 0
///
/// Offset: 0x14 (32 bits) or 0x20 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub section_size: u64, pub section_size: u64,
pub section_link: u32, pub section_link: u32,
pub section_info: u32, pub section_info: u32,
/// Contain the required alignment of the section, must be a power of 2
///
/// Offset: 0x20 (32 bits) or 0x30 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub required_align: u64, pub required_align: u64,
/// Contain the size of each entry, for sections that contain fixed size entries, otherwise 0
///
/// Offset: 0x24 (32 bits) or 0x38 (64 bits), size: 4 (32 bits) or 8 (64 bits) bytes
pub entry_size: u64 pub entry_size: u64
} }
@ -219,20 +348,25 @@ impl SectionHeader {
self.flags & key as u64 != 0 self.flags & key as u64 != 0
} }
fn get_name_offset(instructions: &Vec<u8>, address: usize) -> Option<u32> { /// Return the offset to a string in .shstrtab that represents the name of this section
fn get_name_offset(instructions: &[u8], address: usize) -> Option<u32> {
get_address_point(instructions, address, true).map(|v| { v as u32 }) get_address_point(instructions, address, true).map(|v| { v as u32 })
// set true to return a u32 // set true to return a u32
} }
fn get_header_type(instructions: &Vec<u8>, address: usize) -> Option<u32> { /// Return the type of header of the section
fn get_header_type(instructions: &[u8], address: usize) -> Option<u32> {
get_address_point(instructions, address + 0x4, true).map(|v| { v as u32 }) get_address_point(instructions, address + 0x4, true).map(|v| { v as u32 })
} }
fn get_flags(instructions: &Vec<u8>, address: usize, is_32bits: bool) -> Option<u64> { /// Return the flags of the section, can hold multiples values, see [`FlagValue`]
fn get_flags(instructions: &[u8], address: usize, is_32bits: bool) -> Option<u64> {
get_address_point(instructions, address + 0x8, is_32bits) get_address_point(instructions, address + 0x8, is_32bits)
} }
fn get_virtual_address(instructions: &Vec<u8>, address: usize, is_32bits: bool) -> Option<u64> {
/// Return the virtual address of the section in memory if the sectino is loaded(see section flag), otherwise 0
fn get_virtual_address(instructions: &[u8], address: usize, is_32bits: bool) -> Option<u64> {
get_address_point(instructions, address + if is_32bits { 0x0C } else { 0x10 }, is_32bits) get_address_point(instructions, address + if is_32bits { 0x0C } else { 0x10 }, is_32bits)
} }
@ -398,7 +532,7 @@ impl Loader {
/// return the memory address of something stored at address /// return the memory address of something stored at address
/// Can return None if the file is smaller than adress + 3 (or 7 if 64 bits), in this case, the elf header is incorrect /// Can return None if the file is smaller than adress + 3 (or 7 if 64 bits), in this case, the elf header is incorrect
fn get_address_point(instructions: &Vec<u8>, address: usize, is_32bits: bool) -> Option<u64> { fn get_address_point(instructions: &[u8], address: usize, is_32bits: bool) -> Option<u64> {
if is_32bits { if is_32bits {
let mut bytes: [u8; 4] = [0; 4]; let mut bytes: [u8; 4] = [0; 4];
bytes[0] = instructions.get(address).copied()?; bytes[0] = instructions.get(address).copied()?;
@ -431,10 +565,10 @@ mod test {
let mut machine = Machine::init_machine(); let mut machine = Machine::init_machine();
let loader = Loader::load_and_parse("./test/riscv_instructions/simple_arithmetics/unsigned_addition").expect("IO Error"); let loader = Loader::load_and_parse("./test/riscv_instructions/simple_arithmetics/unsigned_addition").expect("IO Error");
loader.load_into_machine(&mut machine, 0).expect("Parsing error"); loader.load_into_machine(&mut machine, 0).expect("Parsing error");
assert_eq!(false, loader.elf_header.is_32bits); assert!(!loader.elf_header.is_32bits);
assert_eq!(false, loader.elf_header.endianess); assert!(!loader.elf_header.endianess);
assert_eq!(true, loader.elf_header.sys_v_abi); assert!(loader.elf_header.sys_v_abi);
assert_eq!(true, loader.elf_header.is_riscv_target); assert!(loader.elf_header.is_riscv_target);
assert_eq!(1, loader.elf_header.version); assert_eq!(1, loader.elf_header.version);
assert_eq!(0x4000, loader.elf_header.entrypoint); assert_eq!(0x4000, loader.elf_header.entrypoint);
assert_eq!(64, loader.elf_header.elf_header_size); assert_eq!(64, loader.elf_header.elf_header_size);