Add elf header parsing

src/simulator/loader.rs

@@ -1,40 +1,11 @@
 use crate::Machine;
 
+use std::f32::consts::E;
 use std::fs;
 use std::io;
 use std::io::BufRead;
 use std::io::Read;
 
-
-
-	/// Load a file into a new machine
-	/// 
-	/// `panic!` when size is not 1, 2, 4 or 8
-    /// `panic!` when the text does not represents instructions in hexadecimal
-	/// 
-	/// ### Parameters
-	/// 
-	/// - **path** the path of the file to load
-	/// - **size** the number of bytes to write (1, 2, 4 or 8)
-#[deprecated]
-pub fn _load(path : &str, instruction_size: i32) -> Machine {
-    let file = fs::File::open(path).expect("Wrong filename");
-    let reader = io::BufReader::new(file);
-    let mut machine = Machine::init_machine();
-
-    for (i,line) in reader.lines().enumerate() {
-        let res = u64::from_str_radix(&line.unwrap(), 16);
-        match res {
-            Ok(value) => {
-                Machine::write_memory(&mut machine, instruction_size, i*instruction_size as usize, value);
-            },
-            _ => panic!()
-        }
-    }
-    println!("{:x}", Machine::read_memory(& mut machine, 4, 0));
-    machine
-}
-
 /// load a 32-bits binary file into the machine
 /// 
 /// ### Parameters
@@ -63,3 +34,170 @@ pub fn load(path: &str, machine: &mut Machine, start_index: usize) -> Result<(),
     // println!("{:04x?}", instructions); // only print loaded program in debug build
     Ok(())
 }
+
+struct Loader {
+    bytes: Vec<u8>
+}
+
+impl Loader {
+
+    pub fn load(path: &str, machine: &mut Machine, start_index: usize) -> Result<Self, std::io::Error> {
+        let mut file = fs::File::open(path)?;
+        let mut instructions: Vec<u8> = Default::default();
+        loop {
+            let mut buf: [u8; 1] = [0; 1];
+            let res = file.read(&mut buf)?;
+            if res == 0 {
+                break; // eof
+            } else {
+                instructions.push(buf[0]);
+            }
+        }
+        // #[cfg(debug_assertions)]
+        // println!("{:04x?}", instructions); // only print loaded program in debug build
+        Ok(Self { bytes: instructions })
+    }
+
+    fn parse(&mut self) -> Result<(), ()> {
+        todo!();
+        Ok(())
+    }
+
+    /// return true if the 4 first bytes constitude the elf magic number
+    fn is_elf(&self) -> bool {
+        self.bytes.get(0..4) == Option::Some(&[0x7f, 0x45, 0x4c, 0x46])
+    }
+
+    /// return true if big endian, false otherwise
+    fn check_endianess(&self) -> bool {
+        self.bytes.get(5) == Option::Some(&2)
+    }
+
+    /// return true if file is 32 bits, false if 64 bits
+    fn is_32bits(&self) -> bool {
+        self.bytes.get(4) == Option::Some(&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
+    fn get_version(&self) -> Option<u8> {
+        self.bytes.get(6).copied() // work as primitives implements Copy
+    }
+
+    /// return true if target abi of the binary file is System V, false otherwise
+    fn is_system_v_elf(&self) -> bool{
+        self.bytes.get(7) == Option::Some(&0)
+    }
+
+    /// return true if specified target instruction set architecture is RISCV
+    fn is_riscv_isa(&self) -> bool {
+        self.bytes.get(0x12) == Option::Some(&0xf7)
+    }
+
+    /// memory address of the entry point from where the process starts its execution
+    fn get_entrypoint(&self, is_32bits: bool) -> Option<u64> {
+        if is_32bits {
+            self.get_address_point(0x18, true)
+        } else {
+            self.get_address_point(0x18, false)
+        }
+    }
+
+    /// Memory address of the start of the program header table
+    fn get_program_header_table_location(&self, is_32bits: bool) -> Option<u64> {
+        if is_32bits {
+            self.get_address_point(0x1c, true)
+        } else {
+            self.get_address_point(0x20, false)
+        }
+    }
+
+    /// Memory address of the start of the section header table 
+    fn get_section_header_table_location(&self, is_32bits: bool) -> Option<u64> {
+        if is_32bits {
+            self.get_address_point(0x20, true)
+        } else {
+            self.get_address_point(0x28, false)
+        }
+    }
+
+    /// Return the size of the header, normally, 0x40 for 64 bits bin and 0x34 for 32 bits
+    fn get_elf_header_size(&self, is_32bits: bool) -> Option<u16> {
+        let address = if is_32bits { 0x28 } else { 0x34 };
+        self.get_u16_value(address)
+    }
+
+    /// return the size of the program header
+    fn get_program_header_size(&self, is_32bits: bool) -> Option<u16> {
+        let address = if is_32bits { 0x2a } else { 0x34 };
+        self.get_u16_value(address)
+    }
+
+    /// return the number of entries in the program header
+    fn get_number_entries_program_header(&self, is_32bits: bool) -> Option<u16> {
+        let address = if is_32bits { 0x2c } else { 0x38 };
+        self.get_u16_value(address)
+    }
+
+    /// Return the size of the section header
+    fn get_section_header_size(&self, is_32bits: bool) -> Option<u16> {
+        let address = if is_32bits { 0x2e } else { 0x3a };
+        self.get_u16_value(address)
+    }
+
+    /// Return the number of entries in the section header
+    fn get_section_header_num_entries(&self, is_32bits: bool) -> Option<u16> {
+        let address = if is_32bits { 0x30 } else { 0x3c };
+        self.get_u16_value(address)
+    }
+
+    /// Return a u16 value, usually for the size or the number of entries inside a header
+    fn get_u16_value(&self, address: usize) -> Option<u16> {
+        let mut bytes: [u8; 2] = [0; 2];
+        bytes[0] = self.bytes.get(address).copied()?;
+        bytes[1] = self.bytes.get(address + 1).copied()?;
+        Option::Some(u16::from_le_bytes(bytes))
+    }
+
+    /// return the memory address of something stored at address
+    /// Can return None if the file is smaller than adress + 4 (or 7 if 64 bits), in this case, the elf header is incorrect
+    fn get_address_point(&self, address: usize, is_32bits: bool) -> Option<u64> {
+        if is_32bits {
+            let mut bytes: [u8; 4] = [0; 4];
+            bytes[0] = self.bytes.get(address).copied()?;
+            bytes[0] = self.bytes.get(address + 1).copied()?;
+            bytes[0] = self.bytes.get(address + 2).copied()?;
+            bytes[0] = self.bytes.get(address + 3).copied()?;
+            Option::Some(u32::from_le_bytes(bytes) as u64)
+        } else {
+            let mut bytes: [u8; 8] = [0; 8];
+            bytes[0] = self.bytes.get(address).copied()?;
+            bytes[0] = self.bytes.get(address + 1).copied()?;
+            bytes[0] = self.bytes.get(address + 2).copied()?;
+            bytes[0] = self.bytes.get(address + 3).copied()?;
+            bytes[0] = self.bytes.get(address + 4).copied()?;
+            bytes[0] = self.bytes.get(address + 5).copied()?;
+            bytes[0] = self.bytes.get(address + 6).copied()?;
+            bytes[0] = self.bytes.get(address + 7).copied()?;
+            Option::Some(u64::from_le_bytes(bytes))
+        }
+    }
+
+
+}
+
+#[cfg(test)]
+mod test {
+    use crate::simulator::{loader::Loader, machine::Machine};
+
+
+    #[test]
+    fn test_parse() {
+        let mut machine = Machine::init_machine();
+        let loader = Loader::load("./test/riscv_instructions/simple_arithmetics/unsigned_addition", &mut machine, 0).expect("IO Error");
+        assert_eq!(true, loader.is_elf());
+        assert_eq!(false, loader.is_32bits());
+        assert_eq!(false, loader.check_endianess());
+    }
+
+}