From a466dd8f2bae702f443cc49aee3f4c3f26878d4b Mon Sep 17 00:00:00 2001
From: Ayke van Laethem <aykevanlaethem@gmail.com>
Date: Mon, 11 Mar 2019 16:18:09 +0100
Subject: [PATCH] main: include .data section in .hex file

The function extracting the firmware image for .hex and .bin files
wasn't working correctly: it only extracted the .text segment and not
the .data segment.
This commit fixes this issue, so that it behaves (hopefully) just like
objcopy -O{ihex|binary}.

Another small change is that the formatting of the .hex file was made
more like the output of objcopy: no entry addres (old Intel CPU
holdover) and 16 bytes of data on each line.
---
 objcopy.go | 75 ++++++++++++++++++++++++++++++++++++++++++++----------
 uf2.go     |  2 +-
 2 files changed, 63 insertions(+), 14 deletions(-)

diff --git a/objcopy.go b/objcopy.go
index d67bcab5..073fb720 100644
--- a/objcopy.go
+++ b/objcopy.go
@@ -2,8 +2,10 @@ package main
 
 import (
 	"debug/elf"
+	"io/ioutil"
 	"os"
 	"path/filepath"
+	"sort"
 
 	"github.com/marcinbor85/gohex"
 )
@@ -21,24 +23,72 @@ func (e ObjcopyError) Error() string {
 	return e.Op + ": " + e.Err.Error()
 }
 
-// ExtractTextSegment returns the .text segment and the first address from the
-// ELF file in the given path.
-func ExtractTextSegment(path string) (uint64, []byte, error) {
+type ProgSlice []*elf.Prog
+
+func (s ProgSlice) Len() int           { return len(s) }
+func (s ProgSlice) Less(i, j int) bool { return s[i].Paddr < s[j].Paddr }
+func (s ProgSlice) Swap(i, j int)      { s[i], s[j] = s[j], s[i] }
+
+// ExtractROM extracts a firmware image and the first load address from the
+// given ELF file. It tries to emulate the behavior of objcopy.
+func ExtractROM(path string) (uint64, []byte, error) {
 	f, err := elf.Open(path)
 	if err != nil {
 		return 0, nil, ObjcopyError{"failed to open ELF file to extract text segment", err}
 	}
 	defer f.Close()
 
-	text := f.Section(".text")
-	if text == nil {
-		return 0, nil, ObjcopyError{"file does not contain .text segment: " + path, nil}
+	// The GNU objcopy command does the following for firmware extraction (from
+	// the man page):
+	// > When objcopy generates a raw binary file, it will essentially produce a
+	// > memory dump of the contents of the input object file. All symbols and
+	// > relocation information will be discarded. The memory dump will start at
+	// > the load address of the lowest section copied into the output file.
+
+	// Find the lowest section address.
+	startAddr := ^uint64(0)
+	for _, section := range f.Sections {
+		if section.Type != elf.SHT_PROGBITS || section.Flags&elf.SHF_ALLOC == 0 {
+			continue
+		}
+		if section.Addr < startAddr {
+			startAddr = section.Addr
+		}
 	}
-	data, err := text.Data()
-	if err != nil {
-		return 0, nil, ObjcopyError{"failed to extract .text segment from ELF file", err}
+
+	progs := make(ProgSlice, 0, 2)
+	for _, prog := range f.Progs {
+		if prog.Type != elf.PT_LOAD || prog.Filesz == 0 {
+			continue
+		}
+		progs = append(progs, prog)
+	}
+	if len(progs) == 0 {
+		return 0, nil, ObjcopyError{"file does not contain ROM segments: " + path, nil}
+	}
+	sort.Sort(progs)
+
+	var rom []byte
+	for _, prog := range progs {
+		if prog.Paddr != progs[0].Paddr+uint64(len(rom)) {
+			return 0, nil, ObjcopyError{"ROM segments are non-contiguous: " + path, nil}
+		}
+		data, err := ioutil.ReadAll(prog.Open())
+		if err != nil {
+			return 0, nil, ObjcopyError{"failed to extract segment from ELF file: " + path, err}
+		}
+		rom = append(rom, data...)
+	}
+	if progs[0].Paddr < startAddr {
+		// The lowest memory address is before the first section. This means
+		// that there is some extra data loaded at the start of the image that
+		// should be discarded.
+		// Example: ELF files where .text doesn't start at address 0 because
+		// there is a bootloader at the start.
+		return startAddr, rom[startAddr-progs[0].Paddr:], nil
+	} else {
+		return progs[0].Paddr, rom, nil
 	}
-	return text.Addr, data, nil
 }
 
 // Objcopy converts an ELF file to a different (simpler) output file format:
@@ -51,7 +101,7 @@ func Objcopy(infile, outfile string) error {
 	defer f.Close()
 
 	// Read the .text segment.
-	addr, data, err := ExtractTextSegment(infile)
+	addr, data, err := ExtractROM(infile)
 	if err != nil {
 		return err
 	}
@@ -65,12 +115,11 @@ func Objcopy(infile, outfile string) error {
 		return err
 	case ".hex":
 		mem := gohex.NewMemory()
-		mem.SetStartAddress(uint32(addr)) // ignored in most cases (Intel-specific)
 		err := mem.AddBinary(uint32(addr), data)
 		if err != nil {
 			return ObjcopyError{"failed to create .hex file", err}
 		}
-		mem.DumpIntelHex(f, 32) // TODO: handle error
+		mem.DumpIntelHex(f, 16) // TODO: handle error
 		return nil
 	default:
 		panic("unreachable")
diff --git a/uf2.go b/uf2.go
index 6d76a025..04013317 100644
--- a/uf2.go
+++ b/uf2.go
@@ -15,7 +15,7 @@ import (
 // ConvertELFFileToUF2File converts an ELF file to a UF2 file.
 func ConvertELFFileToUF2File(infile, outfile string) error {
 	// Read the .text segment.
-	_, data, err := ExtractTextSegment(infile)
+	_, data, err := ExtractROM(infile)
 	if err != nil {
 		return err
 	}