DWARF writer in V

dwarfwriter.v

module dwarfwriter

import encoding.binary

// SymEntry represents a symbol entry with address and symbol name
pub struct SymEntry {
pub:
	addr   u64
	symbol string
}

// LineEntry represents a line entry with address, file, and line number
pub struct LineEntry {
pub:
	addr u64
	file string
	line int
}

// Helper functions for appending binary data in little-endian format

fn u16_bytes(val u16) []u8 {
	mut b := []u8{len: 2}
	binary.little_endian_put_u16(mut b, val)
	return b
}

fn u32_bytes(val u32) []u8 {
	mut b := []u8{len: 4}
	binary.little_endian_put_u32(mut b, val)
	return b
}

fn u64_bytes(val u64) []u8 {
	mut b := []u8{len: 8}
	binary.little_endian_put_u64(mut b, val)
	return b
}

fn append_uleb128(mut buf []u8, val u32) {
	mut v := val
	for {
		mut byte := u8(v & 0x7f)
		v >>= 7
		if v != 0 {
			byte |= 0x80
		}
		buf << byte
		if v == 0 {
			break
		}
	}
}

fn append_sleb128(mut buf []u8, delta i64) {
	mut val := u64(delta)
	mut more := true
	for more {
		mut byte := u8(val & 0x7f)
		val >>= 7
		sign := if delta < 0 { 1 } else { 0 }
		if (val == 0 && sign == 0) || (i64(val) == -1 && sign == 1) {
			more = false
		} else {
			byte |= 0x80
		}
		buf << byte
	}
}

fn write_at(mut buf []u8, off int, data []u8) {
	for buf.len <= off + data.len {
		buf << 0
	}
	for i, b in data {
		buf[off + i] = b
	}
}

// create_elf_with_dwarf creates an ELF 64-bit object with DWARF debug info
pub fn create_elf_with_dwarf(lines []LineEntry, symbols []SymEntry) []u8 {
	mut buf := []u8{}
	mut text_off := 0
	mut text_sz := 0
	mut debug_info_start := 0
	mut debug_abbrev_start := 0
	mut debug_line_start := 0
	mut symtab_off := 0
	mut symtab_sz := u32(0)
	mut strtab_off := 0
	mut strtab_sz := u32(0)
	mut shoff_off := 0
	mut shdrs_off := 0

	// ELF header
	buf << '\x7fELF'.bytes() // e_ident[EI_MAG]
	buf << u8(2)             // e_ident[EI_CLASS] = ELFCLASS64
	buf << u8(1)             // e_ident[EI_DATA] = ELFDATA2LSB
	buf << u8(1)             // e_ident[EI_VERSION] = EV_CURRENT
	buf << u8(0)             // e_ident[EI_OSABI] = ELFOSABI_NONE
	buf << u8(0)             // e_ident[EI_ABIVERSION]
	for _ in 0..7 {
		buf << u8(0)         // e_ident[EI_PAD]
	}
	buf << u16_bytes(1)      // e_type = ET_REL
	buf << u16_bytes(0x3E)   // e_machine = EM_X86_64
	buf << u32_bytes(1)      // e_version = EV_CURRENT
	buf << u64_bytes(0)      // e_entry
	buf << u64_bytes(0)      // e_phoff
	shoff_off = buf.len
	buf << u64_bytes(0)      // e_shoff (placeholder)
	buf << u32_bytes(0)      // e_flags
	buf << u16_bytes(64)     // e_ehsize
	buf << u16_bytes(0)      // e_phentsize
	buf << u16_bytes(0)      // e_phnum
	buf << u16_bytes(64)     // e_shentsize
	buf << u16_bytes(8)      // e_shnum = 8 (null + 7 sections)
	buf << u16_bytes(7)      // e_shstrndx = 7 (.shstrtab)

	// .text
	text_off = buf.len
	buf << u8(0xC3) // RET
	text_sz = buf.len - text_off

	// .debug_info
	debug_info_start = buf.len
	cu_len_off := buf.len
	buf << u32_bytes(0) // unit_length (placeholder)
	buf << u16_bytes(2) // DWARF version 2
	buf << u32_bytes(0) // debug_abbrev_offset
	buf << u8(8)        // address_size
	buf << u8(0x01)     // Abbrev code 1
	buf << u8(0x0c)     // DW_AT_language (DW_LANG_C99)
	buf << 'main.c'.bytes()
	buf << u8(0)
	buf << '.'.bytes()
	buf << u8(0)
	buf << u64_bytes(0) // low_pc
	buf << u64_bytes(1) // high_pc
	buf << u32_bytes(0) // stmt_list
	for sym in symbols {
		buf << u8(0x02)
		buf << sym.symbol.bytes()
		buf << u8(0)
		buf << u64_bytes(sym.addr)
		buf << u64_bytes(sym.addr + 1)
	}
	buf << u8(0) // end
	cu_end := buf.len
	cu_len := u32(cu_end - (cu_len_off + 4))
	mut b_cu := []u8{len: 4}
	binary.little_endian_put_u32(mut b_cu, cu_len)
	write_at(mut buf, cu_len_off, b_cu)

	// .debug_abbrev
	debug_abbrev_start = buf.len
	buf << u8(0x01) // abbrev code 1
	buf << u8(0x11) // DW_TAG_compile_unit
	buf << u8(0x01) // has children
	buf << u8(0x13)
	buf << u8(0x0b)
	buf << u8(0x03)
	buf << u8(0x08)
	buf << u8(0x1b)
	buf << u8(0x08)
	buf << u8(0x11)
	buf << u8(0x01)
	buf << u8(0x12)
	buf << u8(0x01)
	buf << u8(0x10)
	buf << u8(0x06)
	buf << u8(0x00)
	buf << u8(0x00)
	buf << u8(0x02) // abbrev code 2
	buf << u8(0x2e) // DW_TAG_subprogram
	buf << u8(0x00) // no children
	buf << u8(0x03)
	buf << u8(0x08)
	buf << u8(0x11)
	buf << u8(0x01)
	buf << u8(0x12)
	buf << u8(0x01)
	buf << u8(0x00)
	buf << u8(0x00)
	buf << u8(0x00) // end

	// .debug_line
	debug_line_start = buf.len
	line_len_off := buf.len
	buf << u32_bytes(0) // unit_length (placeholder)
	buf << u16_bytes(2) // version
	hdr_len_off := buf.len
	buf << u32_bytes(0) // header_length (placeholder)
	buf << u8(1)        // minimum_instruction_length
	buf << u8(1)        // default_is_stmt
	buf << u8(0xFB)     // line_base
	buf << u8(14)       // line_range
	buf << u8(13)       // opcode_base
	std_op_len := [u8(0), 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1]
	for len_val in std_op_len {
		buf << len_val
	}
	// dirs (empty)
	buf << u8(0)
	// files
	mut files := []string{}
	for le in lines {
		if le.file !in files {
			files << le.file
		}
	}
	for f in files {
		buf << f.bytes()
		buf << u8(0)
		buf << u8(0) // dir index
		buf << u8(0) // mod time
		buf << u8(0) // file length
	}
	buf << u8(0) // end files
	// line program
	mut prev_line := 1
	mut prev_file_idx := 1
	for le in lines {
		// DW_LNE_set_address
		buf << u8(0)
		buf << u8(1 + 8)
		buf << u8(0x02)
		buf << u64_bytes(le.addr)
		// DW_LNS_set_file
		mut file_idx := 1
		for i, f in files {
			if f == le.file {
				file_idx = i + 1
				break
			}
		}
		if file_idx != prev_file_idx {
			buf << u8(0x04)
			append_uleb128(mut buf, u32(file_idx))
			prev_file_idx = file_idx
		}
		// DW_LNS_advance_line
		buf << u8(0x03)
		delta := i64(le.line) - i64(prev_line)
		append_sleb128(mut buf, delta)
		prev_line = le.line
		// DW_LNS_copy
		buf << u8(0x01)
	}
	// DW_LNE_end_sequence
	buf << u8(0)
	buf << u8(0x01)
	buf << u8(0x01)
	line_ops_end := buf.len
	hdr_len := u32(line_ops_end - (hdr_len_off + 4))
	mut b_hdr := []u8{len: 4}
	binary.little_endian_put_u32(mut b_hdr, hdr_len)
	write_at(mut buf, hdr_len_off, b_hdr)
	line_end := buf.len
	line_len := u32(line_end - (line_len_off + 4))
	mut b_line := []u8{len: 4}
	binary.little_endian_put_u32(mut b_line, line_len)
	write_at(mut buf, line_len_off, b_line)

	// Symbol table
	sym_count := symbols.len
	mut str_offsets := []u32{len: sym_count}
	mut cur_off := u32(1)
	for i, se in symbols {
		str_offsets[i] = cur_off
		cur_off += u32(se.symbol.len + 1)
	}
	symtab_off = buf.len
	// NULL symbol
	buf << u32_bytes(0) // st_name
	buf << u8(0)        // st_info
	buf << u8(0)        // st_other
	buf << u16_bytes(0) // st_shndx
	buf << u64_bytes(0) // st_value
	buf << u64_bytes(0) // st_size
	for i, se in symbols {
		buf << u32_bytes(str_offsets[i]) // st_name
		buf << u8(0x12)                  // st_info (STB_GLOBAL | STT_FUNC)
		buf << u8(0)                     // st_other
		buf << u16_bytes(1)              // st_shndx (.text)
		buf << u64_bytes(se.addr)        // st_value
		buf << u64_bytes(1)              // st_size
	}
	symtab_end := buf.len
	symtab_sz = u32(symtab_end - symtab_off)

	// String table
	strtab_off = buf.len
	buf << u8(0) // empty string
	for se in symbols {
		buf << se.symbol.bytes()
		buf << u8(0)
	}
	strtab_end := buf.len
	strtab_sz = u32(strtab_end - strtab_off)

	// Section header string table
	shstr_off := buf.len
	mut sh_names := []u32{len: 8}
	buf << u8(0) // null
	sh_names[1] = u32(buf.len - shstr_off)
	buf << '.text'.bytes()
	buf << u8(0)
	sh_names[2] = u32(buf.len - shstr_off)
	buf << '.debug_info'.bytes()
	buf << u8(0)
	sh_names[3] = u32(buf.len - shstr_off)
	buf << '.debug_abbrev'.bytes()
	buf << u8(0)
	sh_names[4] = u32(buf.len - shstr_off)
	buf << '.debug_line'.bytes()
	buf << u8(0)
	sh_names[5] = u32(buf.len - shstr_off)
	buf << '.symtab'.bytes()
	buf << u8(0)
	sh_names[6] = u32(buf.len - shstr_off)
	buf << '.strtab'.bytes()
	buf << u8(0)
	sh_names[7] = u32(buf.len - shstr_off)
	buf << '.shstrtab'.bytes()
	buf << u8(0)
	shstr_end := buf.len
	shstr_sz := u32(shstr_end - shstr_off)

	// Patch section header table offset
	shdrs_off = buf.len
	mut b_shoff := []u8{len: 8}
	binary.little_endian_put_u64(mut b_shoff, u64(shdrs_off))
	write_at(mut buf, shoff_off, b_shoff)

	// Section headers
	// NULL section
	for _ in 0..64 {
		buf << u8(0)
	}
	// .text
	buf << u32_bytes(sh_names[1]) // sh_name
	buf << u32_bytes(1)          // sh_type = SHT_PROGBITS
	buf << u64_bytes(6)          // sh_flags = SHF_ALLOC | SHF_EXECINSTR
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(text_off)) // sh_offset
	buf << u64_bytes(u64(text_sz))  // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(16)         // sh_addralign
	buf << u64_bytes(0)          // sh_entsize
	// .debug_info
	buf << u32_bytes(sh_names[2]) // sh_name
	buf << u32_bytes(1)          // sh_type = SHT_PROGBITS
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(debug_info_start)) // sh_offset
	buf << u64_bytes(u64(debug_abbrev_start - debug_info_start)) // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(1)          // sh_addralign
	buf << u64_bytes(0)          // sh_entsize
	// .debug_abbrev
	buf << u32_bytes(sh_names[3]) // sh_name
	buf << u32_bytes(1)          // sh_type = SHT_PROGBITS
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(debug_abbrev_start)) // sh_offset
	buf << u64_bytes(u64(debug_line_start - debug_abbrev_start)) // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(1)          // sh_addralign
	buf << u64_bytes(0)          // sh_entsize
	// .debug_line
	buf << u32_bytes(sh_names[4]) // sh_name
	buf << u32_bytes(1)          // sh_type = SHT_PROGBITS
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(debug_line_start)) // sh_offset
	buf << u64_bytes(u64(symtab_off - debug_line_start)) // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(1)          // sh_addralign
	buf << u64_bytes(0)          // sh_entsize
	// .symtab
	buf << u32_bytes(sh_names[5]) // sh_name
	buf << u32_bytes(2)          // sh_type = SHT_SYMTAB
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(symtab_off)) // sh_offset
	buf << u64_bytes(u64(symtab_sz))  // sh_size
	buf << u32_bytes(6)          // sh_link (.strtab)
	buf << u32_bytes(1)          // sh_info (first non-local symbol)
	buf << u64_bytes(8)          // sh_addralign
	buf << u64_bytes(24)         // sh_entsize
	// .strtab
	buf << u32_bytes(sh_names[6]) // sh_name
	buf << u32_bytes(3)          // sh_type = SHT_STRTAB
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(strtab_off)) // sh_offset
	buf << u64_bytes(u64(strtab_sz))  // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(1)          // sh_addralign
	buf << u64_bytes(0)          // sh_entsize
	// .shstrtab
	buf << u32_bytes(sh_names[7]) // sh_name
	buf << u32_bytes(3)          // sh_type = SHT_STRTAB
	buf << u64_bytes(0)          // sh_flags
	buf << u64_bytes(0)          // sh_addr
	buf << u64_bytes(u64(shstr_off)) // sh_offset
	buf << u64_bytes(u64(shstr_sz))  // sh_size
	buf << u32_bytes(0)          // sh_link
	buf << u32_bytes(0)          // sh_info
	buf << u64_bytes(1)          // sh_addralign
	buf << u64_bytes(0)          // sh_entsize

	return buf
}

test_dwarf.v

import dwarfwriter
import os

fn main() {
	mut lines := []dwarfwriter.LineEntry{}
	lines << dwarfwriter.LineEntry{
		addr: 0x1000
		file: 'main.c'
		line: 42
	}
	lines << dwarfwriter.LineEntry{
		addr: 0x2000
		file: 'main.c'
		line: 53
	}
	lines << dwarfwriter.LineEntry{
		addr: 0x3000
		file: 'foo.c'
		line: 63
	}
	mut symbols := []dwarfwriter.SymEntry{}
	symbols << dwarfwriter.SymEntry{
		addr: 0x1000
		symbol: 'main'
	}
	symbols << dwarfwriter.SymEntry{
		addr: 0x2000
		symbol: 'check'
	}

	elf_buf := dwarfwriter.create_elf_with_dwarf(lines, symbols)
	println('Generated ELF buffer of size ${elf_buf.len}')
	os.write_bytes('dwarf.elf', elf_buf)!
}