[Mach-O] Handle .bss

macho/input-sections.cc

@@ -13,6 +13,7 @@ std::ostream &operator<<(std::ostream &out, const InputSection &sec) {
 InputSection::InputSection(Context &ctx, ObjectFile &file, const MachSection &hdr)
   : file(file), hdr(hdr),
     osec(*OutputSection::get_instance(ctx, hdr.get_segname(), hdr.get_sectname())) {
+  if (hdr.type != S_ZEROFILL)
     contents = file.mf->get_contents().substr(hdr.offset, hdr.size);
   subsections.push_back({*this, 0, (u32)contents.size(), (u32)hdr.addr});
 }

macho/main.cc

@@ -62,6 +62,9 @@ static void create_synthetic_chunks(Context &ctx) {
   ctx.data_seg.chunks.push_back(&ctx.lazy_symbol_ptr);
   ctx.data_seg.chunks.push_back(ctx.data);
 
+  if (!ctx.bss->members.empty())
+    ctx.data_seg.chunks.push_back(ctx.bss);
+
   ctx.linkedit_seg.chunks.push_back(&ctx.rebase);
   ctx.linkedit_seg.chunks.push_back(&ctx.bind);
   ctx.linkedit_seg.chunks.push_back(&ctx.lazy_bind);

macho/mold.h

@@ -127,6 +127,7 @@ std::ostream &operator<<(std::ostream &out, const InputSection &sec);
 class Subsection {
 public:
   std::string_view get_contents() {
+    assert(isec.hdr.type != S_ZEROFILL);
     return isec.contents.substr(input_offset, input_size);
   }
 
@@ -583,8 +584,11 @@ struct Context {
   Context() {
     text = OutputSection::get_instance(*this, "__TEXT", "__text");
     data = OutputSection::get_instance(*this, "__DATA", "__data");
+    bss = OutputSection::get_instance(*this, "__DATA", "__bss");
     cstring = OutputSection::get_instance(*this, "__TEXT", "__cstring");
     common = OutputSection::get_instance(*this, "__TEXT", "__common");
+
+    bss->hdr.type = S_ZEROFILL;
     cstring->hdr.type = S_CSTRING_LITERALS;
   }
 
@@ -660,6 +664,7 @@ struct Context {
 
   OutputSection *text = nullptr;
   OutputSection *data = nullptr;
+  OutputSection *bss = nullptr;
   OutputSection *cstring = nullptr;
   OutputSection *common = nullptr;
 

macho/output-chunks.cc

@@ -292,11 +292,13 @@ void OutputSection::copy_buf(Context &ctx) {
   u8 *buf = ctx.buf + hdr.offset;
 
   for (Subsection *subsec : members) {
+    if (subsec->isec.hdr.type != S_ZEROFILL) {
       std::string_view data = subsec->get_contents();
       memcpy(buf + subsec->output_offset, data.data(), data.size());
       subsec->apply_reloc(ctx, buf + subsec->output_offset);
     }
   }
+}
 
 OutputSegment::OutputSegment(std::string_view name, u32 prot, u32 flags) {
   cmd.cmd = LC_SEGMENT_64;
@@ -310,19 +312,36 @@ void OutputSegment::set_offset(Context &ctx, i64 fileoff, u64 vmaddr) {
   cmd.fileoff = fileoff;
   cmd.vmaddr = vmaddr;
 
-  i64 offset = 0;
+  i64 i = 0;
 
-  for (Chunk *sec : chunks) {
-    offset = align_to(offset, 1 << sec->hdr.p2align);
-    sec->hdr.addr = vmaddr + offset;
-    sec->hdr.offset = fileoff + offset;
-    sec->compute_size(ctx);
-    offset += sec->hdr.size;
+  while (i < chunks.size() && chunks[i]->hdr.type != S_ZEROFILL) {
+    Chunk &sec = *chunks[i++];
+    fileoff = align_to(fileoff, 1 << sec.hdr.p2align);
+    vmaddr = align_to(vmaddr, 1 << sec.hdr.p2align);
+
+    sec.hdr.offset = fileoff;
+    sec.hdr.addr = vmaddr;
+
+    sec.compute_size(ctx);
+    fileoff += sec.hdr.size;
+    vmaddr += sec.hdr.size;
   }
 
-  cmd.vmsize = align_to(offset, PAGE_SIZE);
-  cmd.filesize =
-    (this == ctx.segments.back()) ? offset : align_to(offset, PAGE_SIZE);
+  while (i < chunks.size()) {
+    Chunk &sec = *chunks[i++];
+    assert(sec.hdr.type == S_ZEROFILL);
+    vmaddr = align_to(vmaddr, 1 << sec.hdr.p2align);
+    sec.hdr.addr = vmaddr;
+    sec.compute_size(ctx);
+    vmaddr += sec.hdr.size;
+  }
+
+  cmd.vmsize = align_to(vmaddr - cmd.vmaddr, PAGE_SIZE);
+
+  if (this == ctx.segments.back())
+    cmd.filesize = fileoff - cmd.fileoff;
+  else
+    cmd.filesize = align_to(fileoff - cmd.fileoff, PAGE_SIZE);
 }
 
 void OutputSegment::copy_buf(Context &ctx) {

test/macho/bss.sh

@@ -0,0 +1,23 @@
+#!/bin/bash
+set -e
+cd $(dirname $0)
+mold=`pwd`/../../ld64.mold
+echo -n "Testing $(basename -s .sh $0) ... "
+t=$(pwd)/../../out/test/macho/$(basename -s .sh $0)
+mkdir -p $t
+
+cat <<EOF | cc -o $t/a.o -c -xc -
+#include <stdio.h>
+
+static int foo[100];
+
+int main() {
+  foo[1] = 5;
+  printf("%d %d %p\n", foo[0], foo[1], foo);
+}
+EOF
+
+clang -fuse-ld=$mold -o $t/exe $t/a.o
+$t/exe | grep -q '^0 5 '
+
+echo OK