[mach-o] Add support and test cases for parsing tentative definitions

llvm-svn: 208919

8 files changed, 245 insertions, 4 deletions

diff --git a/lld/lib/ReaderWriter/MachO/Atoms.h b/lld/lib/ReaderWriter/MachO/Atoms.h
index e4bffce5a8b..b21a1ac5005 100644
--- a/lld/lib/ReaderWriter/MachO/Atoms.h
+++ b/lld/lib/ReaderWriter/MachO/Atoms.h
@@ -36,6 +36,37 @@ private:
   const ArrayRef<uint8_t> _content;
   const Scope _scope;
 };
+
+
+class MachOTentativeDefAtom : public SimpleDefinedAtom {
+public:
+  MachOTentativeDefAtom(const File &f, const StringRef name, Scope scope,
+                        uint64_t size, DefinedAtom::Alignment align)
+      : SimpleDefinedAtom(f), _name(name), _scope(scope), _size(size),
+        _align(align) {}
+
+  uint64_t size() const override { return _size; }
+
+  Merge merge() const override { return DefinedAtom::mergeAsTentative; }
+
+  ContentType contentType() const override { return DefinedAtom::typeZeroFill; }
+
+  Alignment alignment() const override { return _align; }
+
+  StringRef name() const override { return _name; }
+
+  Scope scope() const override { return _scope; }
+
+  ArrayRef<uint8_t> rawContent() const override { return ArrayRef<uint8_t>(); }
+
+private:
+  const StringRef _name;
+  const Scope _scope;
+  const uint64_t _size;
+  const DefinedAtom::Alignment _align;
+};
+
+
 } // mach_o
 } // lld
 
diff --git a/lld/lib/ReaderWriter/MachO/File.h b/lld/lib/ReaderWriter/MachO/File.h
index 68aea4a042c..767d5230cf6 100644
--- a/lld/lib/ReaderWriter/MachO/File.h
+++ b/lld/lib/ReaderWriter/MachO/File.h
@@ -43,6 +43,18 @@ public:
     addAtom(*atom);
   }
 
+  void addTentativeDefAtom(StringRef name, Atom::Scope scope, uint64_t size,
+                           DefinedAtom::Alignment align, bool copyRefs) {
+    if (copyRefs) {
+      // Make a copy of the atom's name that is owned by this file.
+      name = name.copy(_allocator);
+    }
+    MachOTentativeDefAtom *atom =
+        new (_allocator) MachOTentativeDefAtom(*this, name, scope, size, align);
+    addAtom(*atom);
+  }
+
+
 private:
   llvm::BumpPtrAllocator _allocator;
 };
diff --git a/lld/lib/ReaderWriter/MachO/MachONormalizedFileBinaryWriter.cpp b/lld/lib/ReaderWriter/MachO/MachONormalizedFileBinaryWriter.cpp
index 1ffb4294c92..10123196a57 100644
--- a/lld/lib/ReaderWriter/MachO/MachONormalizedFileBinaryWriter.cpp
+++ b/lld/lib/ReaderWriter/MachO/MachONormalizedFileBinaryWriter.cpp
@@ -726,6 +726,8 @@ error_code MachOFileLayout::writeLoadCommands() {
 void MachOFileLayout::writeSectionContent() {
   for (const Section &s : _file.sections) {
     // Copy all section content to output buffer.
+    if (s.type == llvm::MachO::S_ZEROFILL)
+      continue;
     uint32_t offset = _sectInfo[&s].fileOffset;
     uint8_t *p = &_buffer[offset];
     memcpy(p, &s.content[0], s.content.size());
diff --git a/lld/lib/ReaderWriter/MachO/MachONormalizedFileFromAtoms.cpp b/lld/lib/ReaderWriter/MachO/MachONormalizedFileFromAtoms.cpp
index b233741e5fb..b5b571898ee 100644
--- a/lld/lib/ReaderWriter/MachO/MachONormalizedFileFromAtoms.cpp
+++ b/lld/lib/ReaderWriter/MachO/MachONormalizedFileFromAtoms.cpp
@@ -176,6 +176,9 @@ SectionInfo *Util::makeSection(DefinedAtom::ContentType type) {
   case DefinedAtom::typeGOT:
      return new (_allocator) SectionInfo("__DATA", "__got",
                             S_NON_LAZY_SYMBOL_POINTERS);
+  case DefinedAtom::typeZeroFill:
+     return new (_allocator) SectionInfo("__DATA", "__bss",
+                            S_ZEROFILL);
   default:
     llvm_unreachable("TO DO: add support for more sections");
     break;
@@ -420,7 +423,8 @@ void Util::appendSection(SectionInfo *si, NormalizedFile &file) {
   // Record where normalized section is.
   si->normalizedSectionIndex = file.sections.size()-1;
   // Copy content from atoms to content buffer for section.
-  // FIXME: zerofill atoms/sections should not take up content space.
+  if (si->type == llvm::MachO::S_ZEROFILL)
+    return;
   uint8_t *sectionContent = file.ownedAllocations.Allocate<uint8_t>(si->size);
   normSect->content = llvm::makeArrayRef(sectionContent, si->size);
   for (AtomInfo &ai : si->atomsAndOffsets) {
diff --git a/lld/lib/ReaderWriter/MachO/MachONormalizedFileToAtoms.cpp b/lld/lib/ReaderWriter/MachO/MachONormalizedFileToAtoms.cpp
index c84ca77367b..59452fbc3dd 100644
--- a/lld/lib/ReaderWriter/MachO/MachONormalizedFileToAtoms.cpp
+++ b/lld/lib/ReaderWriter/MachO/MachONormalizedFileToAtoms.cpp
@@ -85,6 +85,19 @@ static void processSymbol(const NormalizedFile &normalizedFile, MachOFile &file,
   file.addDefinedAtom(sym.name, atomContent, atomScope(sym.scope), copyRefs);
 }
 
+
+static void processUndefindeSymbol(MachOFile &file, const Symbol &sym,
+                                  bool copyRefs) {
+  // Undefinded symbols with n_value!=0 are actually tentative definitions.
+  if (sym.value == Hex64(0)) {
+    file.addUndefinedAtom(sym.name, copyRefs);
+  } else {
+    file.addTentativeDefAtom(sym.name, atomScope(sym.scope), sym.value,
+                              DefinedAtom::Alignment(sym.desc >> 8), copyRefs);
+  }
+}
+
+
 static ErrorOr<std::unique_ptr<lld::File>>
 normalizedObjectToAtoms(const NormalizedFile &normalizedFile, StringRef path,
                         bool copyRefs) {
@@ -99,7 +112,7 @@ normalizedObjectToAtoms(const NormalizedFile &normalizedFile, StringRef path,
   }
 
   for (auto &sym : normalizedFile.undefinedSymbols) {
-    file->addUndefinedAtom(sym.name, copyRefs);
+    processUndefindeSymbol(*file, sym, copyRefs);
   }
 
   return std::unique_ptr<File>(std::move(file));
diff --git a/lld/lib/ReaderWriter/MachO/MachONormalizedFileYAML.cpp b/lld/lib/ReaderWriter/MachO/MachONormalizedFileYAML.cpp
index d3377399c2e..fb63cac227e 100644
--- a/lld/lib/ReaderWriter/MachO/MachONormalizedFileYAML.cpp
+++ b/lld/lib/ReaderWriter/MachO/MachONormalizedFileYAML.cpp
@@ -406,8 +406,18 @@ struct MappingTraits<Symbol> {
     io.mapRequired("type",    sym.type);
     io.mapOptional("scope",   sym.scope, SymbolScope(0));
     io.mapOptional("sect",    sym.sect, (uint8_t)0);
-    io.mapOptional("desc",    sym.desc, SymbolDesc(0));
-    io.mapRequired("value",   sym.value);
+    if (sym.type == llvm::MachO::N_UNDF) {
+      // In undef symbols, desc field contains alignment/ordinal info
+      // which is better represented as a hex vaule.
+      uint16_t t1 = sym.desc;
+      Hex16 t2 = t1;
+      io.mapOptional("desc",  t2, Hex16(0));
+      sym.desc = t2;
+    } else {
+      // In defined symbols, desc fit is a set of option bits.
+      io.mapOptional("desc",    sym.desc, SymbolDesc(0));
+    }
+    io.mapRequired("value",  sym.value);
   }
 };
 
diff --git a/lld/test/mach-o/parse-data.yaml b/lld/test/mach-o/parse-data.yaml
new file mode 100644
index 00000000000..c09162798d2
--- /dev/null
+++ b/lld/test/mach-o/parse-data.yaml
@@ -0,0 +1,81 @@
+# RUN: lld -flavor darwin -arch x86_64 -r -print_atoms %s -o %t  | FileCheck %s
+#
+# Test parsing of mach-o data symbols.
+#
+# long a = 0x0807060504030201;
+# int b = 0x14131211;
+# int c = 0x24232221;
+# static int s1;
+# static int s2 = 0x34333231;
+#
+#
+
+
+--- !mach-o
+arch:            x86_64
+file-type:       MH_OBJECT
+flags:           [ MH_SUBSECTIONS_VIA_SYMBOLS ]
+has-UUID:        false
+OS:              unknown
+sections:
+  - segment:         __DATA
+    section:         __data
+    type:            S_REGULAR
+    attributes:      [  ]
+    alignment:       3
+    address:         0x0000000000000000
+    content:         [ 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08,
+                       0x11, 0x12, 0x13, 0x14, 0x21, 0x22, 0x23, 0x24,
+                       0x31, 0x32, 0x33, 0x34 ]
+  - segment:         __DATA
+    section:         __bss
+    type:            S_ZEROFILL
+    attributes:      [  ]
+    alignment:       2
+    address:         0x0000000000000014
+    content:         [ 0x00, 0x00, 0x00, 0x00 ]
+local-symbols:
+  - name:            _s1
+    type:            N_SECT
+    sect:            2
+    value:           0x0000000000000014
+  - name:            _s2
+    type:            N_SECT
+    sect:            1
+    value:           0x0000000000000010
+global-symbols:
+  - name:            _a
+    type:            N_SECT
+    scope:           [ N_EXT ]
+    sect:            1
+    value:           0x0000000000000000
+  - name:            _b
+    type:            N_SECT
+    scope:           [ N_EXT ]
+    sect:            1
+    value:           0x0000000000000008
+  - name:            _c
+    type:            N_SECT
+    scope:           [ N_EXT ]
+    sect:            1
+    value:           0x000000000000000C
+...
+
+# CHECK: defined-atoms:
+# CHECK:   - name:            _a
+# CHECK:     scope:           global
+# CHECK:     content:         [ 01, 02, 03, 04, 05, 06, 07, 08 ]
+
+# CHECK:   - name:            _b
+# CHECK:     scope:           global
+# CHECK:     content:         [ 11, 12, 13, 14 ]
+
+# CHECK:   - name:            _c
+# CHECK:     scope:           global
+# CHECK:     content:         [ 21, 22, 23, 24 ]
+
+# CHECK:   - name:            _s1
+# CHECK:     content:         [ 00, 00, 00, 00 ]
+
+# CHECK:   - name:            _s2
+# CHECK:     content:         [ 31, 32, 33, 34 ]
diff --git a/lld/test/mach-o/parse-tentative-defs.yaml b/lld/test/mach-o/parse-tentative-defs.yaml
new file mode 100644
index 00000000000..fc75167bd89
--- /dev/null
+++ b/lld/test/mach-o/parse-tentative-defs.yaml
@@ -0,0 +1,88 @@
+# RUN: lld -flavor darwin -arch x86_64 -r -print_atoms %s -o %t  | FileCheck %s
+#
+# Test parsing of tentative definitions, including size, scope, and alignment.
+#
+#
+# int tent4;
+# long tent8;
+# __attribute__((visibility("hidden"))) int tentHidden;
+# __attribute__((aligned(16))) int tent4_16;
+# __attribute__((aligned(32))) long tent64_32[8];
+#
+
+--- !mach-o
+arch:            x86_64
+file-type:       MH_OBJECT
+flags:           [ MH_SUBSECTIONS_VIA_SYMBOLS ]
+has-UUID:        false
+OS:              unknown
+sections:
+  - segment:         __TEXT
+    section:         __tex
+    type:            S_REGULAR
+    attributes:      [ S_ATTR_PURE_INSTRUCTIONS ]
+    address:         0x0000000000000000
+undefined-symbols:
+  - name:            _tent4
+    type:            N_UNDF
+    scope:           [ N_EXT ]
+    desc:            0x0200
+    value:           0x0000000000000004
+  - name:            _tent4_16
+    type:            N_UNDF
+    scope:           [ N_EXT ]
+    desc:            0x0400
+    value:           0x0000000000000004
+  - name:            _tent64_32
+    type:            N_UNDF
+    scope:           [ N_EXT ]
+    desc:            0x0500
+    value:           0x0000000000000040
+  - name:            _tent8
+    type:            N_UNDF
+    scope:           [ N_EXT ]
+    desc:            0x0300
+    value:           0x0000000000000008
+  - name:            _tentHidden
+    type:            N_UNDF
+    scope:           [ N_EXT, N_PEXT ]
+    desc:            0x0200
+    value:           0x0000000000000004
+...
+
+
+# CHECK:    defined-atoms:
+# CHECK:       name:             _tent4
+# CHECK:        scope:           global
+# CHECK:        type:            zero-fill
+# CHECK:        size:            4
+# CHECK:        merge:           as-tentative
+# CHECK:        alignment:       2^2
+
+# CHECK:       name:             _tent4_16
+# CHECK:        scope:           global
+# CHECK:        type:            zero-fill
+# CHECK:        size:            4
+# CHECK:        merge:           as-tentative
+# CHECK:        alignment:       2^4
+
+# CHECK:       name:             _tent64_32
+# CHECK:        scope:           global
+# CHECK:        type:            zero-fill
+# CHECK:        size:            64
+# CHECK:        merge:           as-tentative
+# CHECK:        alignment:       2^5
+
+# CHECK:       name:             _tent8
+# CHECK:        scope:           global
+# CHECK:        type:            zero-fill
+# CHECK:        size:            8
+# CHECK:        merge:           as-tentative
+# CHECK:        alignment:       2^3
+
+# CHECK:       name:             _tentHidden
+# CHECK:        scope:           hidden
+# CHECK:        type:            zero-fill
+# CHECK:        size:            4
+# CHECK:        merge:           as-tentative
+# CHECK:        alignment:       2^2