feat(elf): add no_std ELF64 parser for loading process binaries

crates/elf/src/lib.rs

@@ -0,0 +1,167 @@
+#![no_std]
+
+#[cfg(test)]
+mod tests;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum ElfError {
+    TooSmall,
+    InvalidMagic,
+    Not64Bit,
+    NotLittleEndian,
+    NotExecutable,
+    NoProgramHeaders,
+    HeaderOutOfBounds,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct Segment {
+    pub vaddr: usize,
+    pub file_offset: usize,
+    pub file_size: usize,
+    pub mem_size: usize,
+    pub flags: SegmentFlags,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct SegmentFlags {
+    pub read: bool,
+    pub write: bool,
+    pub execute: bool,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct ElfInfo {
+    pub entry_point: usize,
+    pub segments: [Option<Segment>; MAX_SEGMENTS],
+    pub segment_count: usize,
+}
+
+pub const MAX_SEGMENTS: usize = 8;
+
+const ELF_MAGIC: [u8; 4] = [0x7f, b'E', b'L', b'F'];
+const ELFCLASS64: u8 = 2;
+const ELFDATA2LSB: u8 = 1;
+const ET_EXEC: u16 = 2;
+const PT_LOAD: u32 = 1;
+
+const ELF_HEADER_SIZE: usize = 64;
+const PHDR_SIZE: usize = 56;
+
+pub fn parse(data: &[u8]) -> Result<ElfInfo, ElfError> {
+    if data.len() < ELF_HEADER_SIZE {
+        return Err(ElfError::TooSmall);
+    }
+
+    if data[0..4] != ELF_MAGIC {
+        return Err(ElfError::InvalidMagic);
+    }
+
+    if data[4] != ELFCLASS64 {
+        return Err(ElfError::Not64Bit);
+    }
+
+    if data[5] != ELFDATA2LSB {
+        return Err(ElfError::NotLittleEndian);
+    }
+
+    let e_type = read_u16(data, 16);
+    if e_type != ET_EXEC {
+        return Err(ElfError::NotExecutable);
+    }
+
+    let entry_point = read_u64(data, 24) as usize;
+    let ph_offset = read_u64(data, 32) as usize;
+    let ph_entry_size = read_u16(data, 54) as usize;
+    let ph_count = read_u16(data, 56) as usize;
+
+    if ph_count == 0 {
+        return Err(ElfError::NoProgramHeaders);
+    }
+
+    let mut segments = [None; MAX_SEGMENTS];
+    let mut segment_count = 0;
+
+    for i in 0..ph_count {
+        let offset = ph_offset + i * ph_entry_size;
+        if offset + PHDR_SIZE > data.len() {
+            return Err(ElfError::HeaderOutOfBounds);
+        }
+
+        let p_type = read_u32(data, offset);
+        if p_type != PT_LOAD {
+            continue;
+        }
+
+        if segment_count >= MAX_SEGMENTS {
+            break;
+        }
+
+        let p_flags = read_u32(data, offset + 4);
+        let p_offset = read_u64(data, offset + 8) as usize;
+        let p_vaddr = read_u64(data, offset + 16) as usize;
+        let p_filesz = read_u64(data, offset + 32) as usize;
+        let p_memsz = read_u64(data, offset + 40) as usize;
+
+        segments[segment_count] = Some(Segment {
+            vaddr: p_vaddr,
+            file_offset: p_offset,
+            file_size: p_filesz,
+            mem_size: p_memsz,
+            flags: SegmentFlags {
+                execute: p_flags & 1 != 0,
+                write: p_flags & 2 != 0,
+                read: p_flags & 4 != 0,
+            },
+        });
+        segment_count += 1;
+    }
+
+    Ok(ElfInfo {
+        entry_point,
+        segments,
+        segment_count,
+    })
+}
+
+impl ElfInfo {
+    pub fn segments(&self) -> impl Iterator<Item = &Segment> {
+        self.segments[..self.segment_count]
+            .iter()
+            .filter_map(|s| s.as_ref())
+    }
+
+    pub fn total_pages(&self) -> usize {
+        let mut total = 0;
+        for seg in self.segments() {
+            total += (seg.mem_size + 4095) / 4096;
+        }
+        total
+    }
+}
+
+fn read_u16(data: &[u8], offset: usize) -> u16 {
+    u16::from_le_bytes([data[offset], data[offset + 1]])
+}
+
+fn read_u32(data: &[u8], offset: usize) -> u32 {
+    u32::from_le_bytes([
+        data[offset],
+        data[offset + 1],
+        data[offset + 2],
+        data[offset + 3],
+    ])
+}
+
+fn read_u64(data: &[u8], offset: usize) -> u64 {
+    u64::from_le_bytes([
+        data[offset],
+        data[offset + 1],
+        data[offset + 2],
+        data[offset + 3],
+        data[offset + 4],
+        data[offset + 5],
+        data[offset + 6],
+        data[offset + 7],
+    ])
+}

crates/elf/src/tests.rs

@@ -0,0 +1,132 @@
+use super::*;
+
+fn minimal_elf64() -> [u8; 120] {
+    let mut buf = [0u8; 120];
+
+    // ELF magic
+    buf[0] = 0x7f;
+    buf[1] = b'E';
+    buf[2] = b'L';
+    buf[3] = b'F';
+
+    // EI_CLASS = ELFCLASS64
+    buf[4] = 2;
+    // EI_DATA = ELFDATA2LSB
+    buf[5] = 1;
+    // EI_VERSION
+    buf[6] = 1;
+
+    // e_type = ET_EXEC (2)
+    buf[16] = 2;
+    buf[17] = 0;
+
+    // e_entry = 0x1000
+    let entry: u64 = 0x1000;
+    buf[24..32].copy_from_slice(&entry.to_le_bytes());
+
+    // e_phoff = 64 (right after ELF header)
+    let phoff: u64 = 64;
+    buf[32..40].copy_from_slice(&phoff.to_le_bytes());
+
+    // e_phentsize = 56
+    buf[54] = 56;
+    buf[55] = 0;
+
+    // e_phnum = 1
+    buf[56] = 1;
+    buf[57] = 0;
+
+    // Program header at offset 64:
+    let ph = 64;
+
+    // p_type = PT_LOAD (1)
+    buf[ph] = 1;
+
+    // p_flags = PF_R | PF_X (4 | 1 = 5)
+    buf[ph + 4] = 5;
+
+    // p_offset = 0
+    let p_offset: u64 = 0;
+    buf[ph + 8..ph + 16].copy_from_slice(&p_offset.to_le_bytes());
+
+    // p_vaddr = 0x1000
+    let p_vaddr: u64 = 0x1000;
+    buf[ph + 16..ph + 24].copy_from_slice(&p_vaddr.to_le_bytes());
+
+    // p_filesz = 120
+    let p_filesz: u64 = 120;
+    buf[ph + 32..ph + 40].copy_from_slice(&p_filesz.to_le_bytes());
+
+    // p_memsz = 4096
+    let p_memsz: u64 = 4096;
+    buf[ph + 40..ph + 48].copy_from_slice(&p_memsz.to_le_bytes());
+
+    buf
+}
+
+#[test]
+fn parse_valid_elf64() {
+    let data = minimal_elf64();
+    let info = parse(&data).unwrap();
+
+    assert_eq!(info.entry_point, 0x1000);
+    assert_eq!(info.segment_count, 1);
+
+    let seg = info.segments().next().unwrap();
+    assert_eq!(seg.vaddr, 0x1000);
+    assert_eq!(seg.file_size, 120);
+    assert_eq!(seg.mem_size, 4096);
+    assert!(seg.flags.read);
+    assert!(seg.flags.execute);
+    assert!(!seg.flags.write);
+}
+
+#[test]
+fn parse_too_small_fails() {
+    let data = [0u8; 10];
+    assert_eq!(parse(&data), Err(ElfError::TooSmall));
+}
+
+#[test]
+fn parse_bad_magic_fails() {
+    let mut data = minimal_elf64();
+    data[0] = 0;
+    assert_eq!(parse(&data), Err(ElfError::InvalidMagic));
+}
+
+#[test]
+fn parse_32bit_fails() {
+    let mut data = minimal_elf64();
+    data[4] = 1; // ELFCLASS32
+    assert_eq!(parse(&data), Err(ElfError::Not64Bit));
+}
+
+#[test]
+fn parse_big_endian_fails() {
+    let mut data = minimal_elf64();
+    data[5] = 2; // ELFDATA2MSB
+    assert_eq!(parse(&data), Err(ElfError::NotLittleEndian));
+}
+
+#[test]
+fn parse_shared_object_fails() {
+    let mut data = minimal_elf64();
+    data[16] = 3; // ET_DYN
+    assert_eq!(parse(&data), Err(ElfError::NotExecutable));
+}
+
+#[test]
+fn total_pages_rounds_up() {
+    let data = minimal_elf64();
+    let info = parse(&data).unwrap();
+    assert_eq!(info.total_pages(), 1);
+}
+
+#[test]
+fn skips_non_load_segments() {
+    let mut data = minimal_elf64();
+    // Change p_type to PT_NOTE (4)
+    data[64] = 4;
+    let info = parse(&data).unwrap();
+    assert_eq!(info.segment_count, 0);
+}