From 948d499231f5633c182d3a7f3987b39625043ad0 Mon Sep 17 00:00:00 2001
From: Eduard Urbach <admin@akyoto.dev>
Date: Sat, 13 Jul 2024 14:18:55 +0200
Subject: [PATCH] Improved documentation

---
 src/build/asm/Assembler.go     |  4 ++--
 src/build/asm/RegisterLabel.go |  2 +-
 src/build/config/config.go     | 13 +++++++---
 src/build/elf/ELF.go           | 44 +++++++++++++++++++---------------
 src/build/elf/Padding.go       |  6 +++++
 src/build/elf/elf.md           |  6 ++++-
 6 files changed, 49 insertions(+), 26 deletions(-)
 create mode 100644 src/build/elf/Padding.go

diff --git a/src/build/asm/Assembler.go b/src/build/asm/Assembler.go
index 08efb10..6ad35c8 100644
--- a/src/build/asm/Assembler.go
+++ b/src/build/asm/Assembler.go
@@ -4,12 +4,12 @@ import "maps"
 
 // Assembler contains a list of instructions.
 type Assembler struct {
-	Instructions []Instruction
 	Data         map[string][]byte
+	Instructions []Instruction
 }
 
 // Merge combines the contents of this assembler with another one.
 func (a *Assembler) Merge(b Assembler) {
-	a.Instructions = append(a.Instructions, b.Instructions...)
 	maps.Copy(a.Data, b.Data)
+	a.Instructions = append(a.Instructions, b.Instructions...)
 }
diff --git a/src/build/asm/RegisterLabel.go b/src/build/asm/RegisterLabel.go
index 4eb23a3..f51bf77 100644
--- a/src/build/asm/RegisterLabel.go
+++ b/src/build/asm/RegisterLabel.go
@@ -8,8 +8,8 @@ import (
 
 // RegisterLabel operates with a register and a label.
 type RegisterLabel struct {
-	Register cpu.Register
 	Label    string
+	Register cpu.Register
 }
 
 // String returns a human readable version.
diff --git a/src/build/config/config.go b/src/build/config/config.go
index 4f97a22..0697fa1 100644
--- a/src/build/config/config.go
+++ b/src/build/config/config.go
@@ -1,10 +1,17 @@
 package config
 
 const (
-	MinAddress  = 0x10000
+	// This is the absolute virtual minimum address we can load a program at, see `sysctl vm.mmap_min_addr`.
+	MinAddress = 0x10000
+
+	// The base address is the virtual address for our ELF file.
 	BaseAddress = 0x40 * MinAddress
-	CodeOffset  = 0x80
-	Align       = 0x10
+
+	// The code offset is the offset of the executable machine code within the file.
+	CodeOffset = 64 + 56 + 56
+
+	// Align decides the alignment of the sections and it must be a multiple of the page size.
+	Align = 0x1000
 )
 
 var (
diff --git a/src/build/elf/ELF.go b/src/build/elf/ELF.go
index b6c3e34..a2b1e4a 100644
--- a/src/build/elf/ELF.go
+++ b/src/build/elf/ELF.go
@@ -11,11 +11,12 @@ import (
 // ELF represents an ELF file.
 type ELF struct {
 	Header
-	CodeHeader ProgramHeader
-	PadCode    []byte
-	Code       []byte
-	PadData    []byte
-	Data       []byte
+	CodeHeader  ProgramHeader
+	DataHeader  ProgramHeader
+	CodePadding []byte
+	Code        []byte
+	DataPadding []byte
+	Data        []byte
 }
 
 // New creates a new ELF binary.
@@ -24,7 +25,7 @@ func New(code []byte, data []byte) *ELF {
 	dataPadding := Padding(dataOffset, config.Align)
 	dataOffset += dataPadding
 
-	elf := &ELF{
+	return &ELF{
 		Header: Header{
 			Magic:                       [4]byte{0x7F, 'E', 'L', 'F'},
 			Class:                       2,
@@ -41,7 +42,7 @@ func New(code []byte, data []byte) *ELF {
 			Flags:                       0,
 			Size:                        HeaderSize,
 			ProgramHeaderEntrySize:      ProgramHeaderSize,
-			ProgramHeaderEntryCount:     1,
+			ProgramHeaderEntryCount:     2,
 			SectionHeaderEntrySize:      SectionHeaderSize,
 			SectionHeaderEntryCount:     0,
 			SectionNameStringTableIndex: 0,
@@ -56,25 +57,30 @@ func New(code []byte, data []byte) *ELF {
 			SizeInMemory:    int64(len(code)),
 			Align:           config.Align,
 		},
-		PadCode: bytes.Repeat([]byte{0}, 8),
-		Code:    code,
-		PadData: bytes.Repeat([]byte{0}, int(dataPadding)),
-		Data:    data,
+		DataHeader: ProgramHeader{
+			Type:            ProgramTypeLOAD,
+			Flags:           ProgramFlagsReadable,
+			Offset:          dataOffset,
+			VirtualAddress:  config.BaseAddress + dataOffset,
+			PhysicalAddress: config.BaseAddress + dataOffset,
+			SizeInFile:      int64(len(data)),
+			SizeInMemory:    int64(len(data)),
+			Align:           config.Align,
+		},
+		CodePadding: nil,
+		Code:        code,
+		DataPadding: bytes.Repeat([]byte{0}, int(dataPadding)),
+		Data:        data,
 	}
-
-	return elf
-}
-
-func Padding(n int64, align int64) int64 {
-	return align - (n % align)
 }
 
 // Write writes the ELF64 format to the given writer.
 func (elf *ELF) Write(writer io.Writer) {
 	binary.Write(writer, binary.LittleEndian, &elf.Header)
 	binary.Write(writer, binary.LittleEndian, &elf.CodeHeader)
-	writer.Write(elf.PadCode)
+	binary.Write(writer, binary.LittleEndian, &elf.DataHeader)
+	writer.Write(elf.CodePadding)
 	writer.Write(elf.Code)
-	writer.Write(elf.PadData)
+	writer.Write(elf.DataPadding)
 	writer.Write(elf.Data)
 }
diff --git a/src/build/elf/Padding.go b/src/build/elf/Padding.go
new file mode 100644
index 0000000..5cca58e
--- /dev/null
+++ b/src/build/elf/Padding.go
@@ -0,0 +1,6 @@
+package elf
+
+// Padding calculates the padding needed to align `n` bytes with the specified alignment.
+func Padding(n int64, align int64) int64 {
+	return align - (n % align)
+}
diff --git a/src/build/elf/elf.md b/src/build/elf/elf.md
index 83df35c..c40f005 100644
--- a/src/build/elf/elf.md
+++ b/src/build/elf/elf.md
@@ -23,4 +23,8 @@ Usually, this value is 65536 (0x1000).
 
 ## Initialization in Linux
 
-See `/lib/modules/$(uname -r)/build/arch/x86/include/asm/elf.h`.
+ELF loader:
+https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/binfmt_elf.c
+
+ELF register definitions:
+https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/x86/include/asm/elf.h