[Mach-O] wip

macho/dumper.cc

@@ -84,7 +84,7 @@ void dump_file(std::string path) {
                 << "\n strsize: 0x" << cmd.strsize
                 << "\n";
 
-      MachoSym *syms = (MachoSym *)(buf + cmd.symoff);
+      MachSym *syms = (MachSym *)(buf + cmd.symoff);
       for (i64 j = 0; j < cmd.nsyms; j++) {
         std::cout << " symbol:"
                   << "\n  name: " << (char *)(buf + cmd.stroff + syms[j].stroff)
@@ -184,7 +184,7 @@ void dump_file(std::string path) {
 //        }
 
         if (sec[j].reloff) {
-          MachoRel *rel = (MachoRel *)(buf + sec[j].reloff);
+          MachRel *rel = (MachRel *)(buf + sec[j].reloff);
           for (i64 k = 0; k < sec[j].nreloc; k++) {
             std::cout << "  reloc: "
                       << "\n   offset: 0x" << rel[k].offset

macho/macho.h

@@ -377,7 +377,7 @@ struct DataInCodeEntry {
 };
 
 // This struct is named `n_list` on BSD and macOS.
-struct MachoSym {
+struct MachSym {
   u32 stroff;
   u8 ext : 1;
   u8 type : 3;
@@ -389,7 +389,7 @@ struct MachoSym {
 };
 
 // This struct is named `relocation_info` on BSD and macOS.
-struct MachoRel {
+struct MachRel {
   i32 offset;
   u32 idx : 24;
   u32 is_pcrel : 1;

macho/main.cc

@@ -10,21 +10,34 @@
 
 namespace mold::macho {
 
-static void print_bytes(u8 *buf, i64 size) {
+void create_synthetic_sections(Context &ctx) {
-  if (size == 0) {
+  auto add = [&](auto &chunk) {
-    std::cout << "[]\n";
+    ctx.chunks.push_back(chunk.get());
-    return;
+  };
-  }
 
-  std::cout << "[" << std::setw(2) << std::setfill('0') << (u32)buf[0];
+  add(ctx.mach_hdr = std::make_unique<MachHeaderChunk>());
-  for (i64 i = 1; i < size; i++)
+  add(ctx.load_cmd = std::make_unique<LoadCommandChunk>());
-    std::cout << " " << std::setw(2) << std::setfill('0') << (u32)buf[i];
+  add(ctx.zero_page = std::make_unique<PageZeroChunk>());
-  std::cout << "]\n";
+}
+
+void compute_chunk_sizes(Context &ctx) {
+  for (Chunk *chunk : ctx.chunks)
+    chunk->update_hdr(ctx);
+}
+
+void assign_file_offsets(Context &ctx) {
+  i64 fileoff = 0;
+
+  for (Chunk *chunk : ctx.chunks) {
+    chunk->fileoff = fileoff;
+    fileoff += chunk->size;
+  }
 }
 
 int main(int argc, char **argv) {
   Context ctx;
 
+  // Parse command line arguments
   if (argc == 1) {
     SyncOut(ctx) << "mold macho stub\n";
     exit(0);
@@ -37,25 +50,27 @@ int main(int argc, char **argv) {
     exit(0);
   }
 
-  if (argc != 2)
+  if (std::string_view(argv[1]) == "-out") {
-    Fatal(ctx) << "usage: ld64.mold <output-file>\n";
+    if (argc != 3)
-  ctx.arg.output = argv[1];
+      Fatal(ctx) << "usage: ld64.mold -out <output-file>\n";
+    ctx.arg.output = argv[2];
 
-  ctx.output_file = std::make_unique<OutputFile>(ctx, ctx.arg.output, 1024, 0777);
+    create_synthetic_sections(ctx);
-  ctx.buf = ctx.output_file->buf;
+    compute_chunk_sizes(ctx);
+    assign_file_offsets(ctx);
 
-  MachHeader &hdr = *(MachHeader *)ctx.buf;
+    ctx.output_file =
-  hdr.magic = 0xfeedfacf;
+      std::make_unique<OutputFile>(ctx, ctx.arg.output, 1024, 0777);
-  hdr.cputype = CPU_TYPE_X86_64;
+    ctx.buf = ctx.output_file->buf;
-  hdr.cpusubtype = CPU_SUBTYPE_X86_64_ALL;
-  hdr.filetype = MH_EXECUTE;
-  hdr.ncmds = 0x10;
-  hdr.sizeofcmds = 0x558;
-  hdr.flags = MH_TWOLEVEL | MH_NOUNDEFS | MH_DYLDLINK | MH_PIE;
 
-  ctx.output_file->close(ctx);
+    for (Chunk *chunk : ctx.chunks)
+      chunk->copy_buf(ctx);
 
-  return 0;
+    ctx.output_file->close(ctx);
+    exit(0);
+  }
+
+  Fatal(ctx) << "usage: ld64.mold\n";
 }
 
 }

macho/mold.h

@@ -9,6 +9,59 @@ namespace mold::macho {
 
 struct Context;
 
+//
+// output-chunks.cc
+//
+
+class Chunk {
+public:
+  // There are three types of OutputChunks:
+  //  - HEADER: the ELF, section or segment headers
+  //  - REGULAR: output sections containing input sections
+  //  - SYNTHETIC: linker-synthesized sections such as got or plt
+  enum Kind : u8 { HEADER, REGULAR, SYNTHETIC };
+
+  virtual ~Chunk() = default;
+  virtual void copy_buf(Context &ctx) {}
+  virtual void update_hdr(Context &ctx) {}
+
+  std::string_view name;
+  Kind kind;
+  i64 size = 0;
+  i64 fileoff = 0;
+  std::vector<u8> load_cmd;
+
+protected:
+  Chunk(Kind kind) : kind(kind) {}
+};
+
+class MachHeaderChunk : public Chunk {
+public:
+  MachHeaderChunk() : Chunk(HEADER) {
+    size = sizeof(MachHeader);
+  }
+
+  void copy_buf(Context &ctx) override;
+};
+
+class LoadCommandChunk : public Chunk {
+public:
+  LoadCommandChunk() : Chunk(HEADER) {}
+
+  void update_hdr(Context &ctx) override;
+  void copy_buf(Context &ctx) override;
+
+  i64 ncmds = 0;
+
+private:
+  std::vector<u8> contents;
+};
+
+class PageZeroChunk : public Chunk {
+public:
+  PageZeroChunk();
+};
+
 //
 // output-file.cc
 //
@@ -56,6 +109,11 @@ struct Context {
 
   std::unique_ptr<OutputFile> output_file;
   u8 *buf;
+
+  std::unique_ptr<MachHeaderChunk> mach_hdr;
+  std::unique_ptr<LoadCommandChunk> load_cmd;
+  std::unique_ptr<PageZeroChunk> zero_page;
+  std::vector<Chunk *> chunks;
 };
 
 int main(int argc, char **argv);

macho/output-chunks.cc

@@ -0,0 +1,52 @@
+#include "mold.h"
+
+namespace mold::macho {
+
+void MachHeaderChunk::copy_buf(Context &ctx) {
+  MachHeader &hdr = *(MachHeader *)(ctx.buf + fileoff);
+  memset(&hdr, 0, sizeof(hdr));
+
+  hdr.magic = 0xfeedfacf;
+  hdr.cputype = CPU_TYPE_X86_64;
+  hdr.cpusubtype = CPU_SUBTYPE_X86_64_ALL;
+  hdr.filetype = MH_EXECUTE;
+  hdr.ncmds = ctx.load_cmd->ncmds;
+  hdr.sizeofcmds = ctx.load_cmd->size;
+  hdr.flags = MH_TWOLEVEL | MH_NOUNDEFS | MH_DYLDLINK | MH_PIE;
+}
+
+void LoadCommandChunk::update_hdr(Context &ctx) {
+  size = 0;
+  ncmds = 0;
+
+  for (Chunk *chunk : ctx.chunks) {
+    if (chunk->load_cmd.size() > 0) {
+      size += chunk->load_cmd.size();
+      ncmds++;
+    }
+  }
+}
+
+void LoadCommandChunk::copy_buf(Context &ctx) {
+  u8 *buf = ctx.buf + fileoff;
+  i64 off = 0;
+
+  for (Chunk *chunk : ctx.chunks) {
+    if (chunk->load_cmd.size() > 0) {
+      std::vector<u8> &vec = chunk->load_cmd;
+      memcpy(buf + off, vec.data(), vec.size());
+      off += vec.size();
+    }
+  }
+}
+
+PageZeroChunk::PageZeroChunk() : Chunk(SYNTHETIC) {
+  load_cmd.resize(sizeof(SegmentCommand));
+  SegmentCommand &cmd = *(SegmentCommand *)load_cmd.data();
+  cmd.cmd = LC_SEGMENT_64;
+  cmd.cmdsize = sizeof(cmd);
+  strcpy(cmd.segname, "__PAGEZERO");
+  cmd.vmsize = 0x100000000;
+}
+
+} // namespace mold::macho

test/macho/dump.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+set -e
+cd $(dirname $0)
+mold=`pwd`/../../ld64.mold
+echo -n "Testing $(basename -s .sh $0) ... "
+t=$(pwd)/../../out/test/elf/$(basename -s .sh $0)
+mkdir -p $t
+
+$mold -out $t/exe
+$mold -dump $t/exe > $t/log
+
+grep -q 'magic: 0xfeedfacf' $t/log
+grep -q 'segname: __PAGEZERO' $t/log
+
+echo OK