[PDB] Add char8_t type

[zequanwu]

Summary

As Zach mentioned here: https://github.com/microsoft/microsoft-pdb/issues/51#issuecomment-954876920.

[rnk]

Please add a test that uses llc and llvm-pdbutil to show the roundtripping of this from LLVM IR with debug info to codeview. Otherwise, looks good, thanks!

[zequanwu]

Rebase and add a test to use llc and llvm-pdbutil to test char8_t.

[rnk]

lgtm

[ljmf00]

Thanks for adding this, I ended up forgoting, since I don't use PDB.

lldb/include/lldb/lldb-enumerations.h

752

You forgot to add it to the Python API. It should also be added at the end to avoid user-code breakage. See D116136 . I can rebase my change and correct this :)

[zequanwu]

lldb/include/lldb/lldb-enumerations.h

752

Thanks, it sounds good.

lldb/include/lldb/lldb-enumerations.h

enum BasicType {
   eBasicTypeChar,
   eBasicTypeSignedChar,
   eBasicTypeUnsignedChar,
   eBasicTypeWChar,
   eBasicTypeSignedWChar,
   eBasicTypeUnsignedWChar,
   eBasicTypeChar16,
   eBasicTypeChar32,
+  eBasicTypeChar8,
   eBasicTypeShort,
   eBasicTypeUnsignedShort,
   eBasicTypeInt,
   eBasicTypeUnsignedInt,
   eBasicTypeLong,
   eBasicTypeUnsignedLong,
   eBasicTypeLongLong,
   eBasicTypeUnsignedLongLong,

lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.cpp

case SimpleTypeKind::NarrowCharacter:
     return lldb::eBasicTypeChar;
   case SimpleTypeKind::SignedCharacter:
   case SimpleTypeKind::SByte:
     return lldb::eBasicTypeSignedChar;
   case SimpleTypeKind::Character16:
     return lldb::eBasicTypeChar16;
   case SimpleTypeKind::Character32:
     return lldb::eBasicTypeChar32;
+  case SimpleTypeKind::Character8:
+    return lldb::eBasicTypeChar8;
   case SimpleTypeKind::Complex80:
     return lldb::eBasicTypeLongDoubleComplex;
   case SimpleTypeKind::Complex64:
     return lldb::eBasicTypeDoubleComplex;
   case SimpleTypeKind::Complex32:
     return lldb::eBasicTypeFloatComplex;
   case SimpleTypeKind::Float128:
   case SimpleTypeKind::Float80:
size_t lldb_private::npdb::GetTypeSizeForSimpleKind(SimpleTypeKind kind) {
   case SimpleTypeKind::WideCharacter:
     return 2;
   case SimpleTypeKind::Boolean8:
   case SimpleTypeKind::Byte:
   case SimpleTypeKind::UnsignedCharacter:
   case SimpleTypeKind::NarrowCharacter:
   case SimpleTypeKind::SignedCharacter:
   case SimpleTypeKind::SByte:
+  case SimpleTypeKind::Character8:
     return 1;
   case SimpleTypeKind::Void:
   default:
     return 0;
   }
 }
 
 PdbTypeSymId lldb_private::npdb::GetBestPossibleDecl(PdbTypeSymId id,

lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp

case SimpleTypeKind::NarrowCharacter:
     return "char";
   case SimpleTypeKind::SignedCharacter:
   case SimpleTypeKind::SByte:
     return "signed char";
   case SimpleTypeKind::Character16:
     return "char16_t";
   case SimpleTypeKind::Character32:
     return "char32_t";
+  case SimpleTypeKind::Character8:
+    return "char8_t";
   case SimpleTypeKind::Complex80:
   case SimpleTypeKind::Complex64:
   case SimpleTypeKind::Complex32:
     return "complex";
   case SimpleTypeKind::Float128:
   case SimpleTypeKind::Float80:
     return "long double";
   case SimpleTypeKind::Float64:

llvm/docs/PDB/TpiStream.rst

enum class SimpleTypeKind : uint32_t {
     HResult = 0x0008,       // OLE/COM HRESULT
 
     SignedCharacter = 0x0010,   // 8 bit signed
     UnsignedCharacter = 0x0020, // 8 bit unsigned
     NarrowCharacter = 0x0070,   // really a char
     WideCharacter = 0x0071,     // wide char
     Character16 = 0x007a,       // char16_t
     Character32 = 0x007b,       // char32_t
+    Character8 = 0x007c,       // char8_t
 
     SByte = 0x0068,       // 8 bit signed int
     Byte = 0x0069,        // 8 bit unsigned int
     Int16Short = 0x0011,  // 16 bit signed
     UInt16Short = 0x0021, // 16 bit unsigned
     Int16 = 0x0072,       // 16 bit signed int
     UInt16 = 0x0073,      // 16 bit unsigned int
     Int32Long = 0x0012,   // 32 bit signed

llvm/include/llvm/DebugInfo/CodeView/TypeIndex.h

enum class SimpleTypeKind : uint32_t {
   HResult = 0x0008,       // OLE/COM HRESULT
 
   SignedCharacter = 0x0010,   // 8 bit signed
   UnsignedCharacter = 0x0020, // 8 bit unsigned
   NarrowCharacter = 0x0070,   // really a char
   WideCharacter = 0x0071,     // wide char
   Character16 = 0x007a,       // char16_t
   Character32 = 0x007b,       // char32_t
+  Character8 = 0x007c,       // char8_t
 
   SByte = 0x0068,       // 8 bit signed int
   Byte = 0x0069,        // 8 bit unsigned int
   Int16Short = 0x0011,  // 16 bit signed
   UInt16Short = 0x0021, // 16 bit unsigned
   Int16 = 0x0072,       // 16 bit signed int
   UInt16 = 0x0073,      // 16 bit unsigned int
   Int32Long = 0x0012,   // 32 bit signed

llvm/include/llvm/DebugInfo/PDB/PDBTypes.h

enum class PDB_BuiltinType {
   Currency = 25,
   Date = 26,
   Variant = 27,
   Complex = 28,
   Bitfield = 29,
   BSTR = 30,
   HResult = 31,
   Char16 = 32,
-  Char32 = 33
+  Char32 = 33,
+  Char8  = 34,
 };
 
 /// These values correspond to the flags that can be combined to control the
 /// return of an undecorated name for a C++ decorated name, and are documented
 /// here: https://msdn.microsoft.com/en-us/library/kszfk0fs.aspx
 enum PDB_UndnameFlags : uint32_t {
   Undname_Complete = 0x0,
   Undname_NoLeadingUnderscores = 0x1,

llvm/lib/CodeGen/AsmPrinter/CodeViewDebug.cpp

case dwarf::DW_ATE_unsigned:
     case 2:  STK = SimpleTypeKind::UInt16Short;       break;
     case 4:  STK = SimpleTypeKind::UInt32;            break;
     case 8:  STK = SimpleTypeKind::UInt64Quad;        break;
     case 16: STK = SimpleTypeKind::UInt128Oct;        break;
     }
     break;
   case dwarf::DW_ATE_UTF:
     switch (ByteSize) {
+    case 1: STK = SimpleTypeKind::Character8; break;
     case 2: STK = SimpleTypeKind::Character16; break;
     case 4: STK = SimpleTypeKind::Character32; break;
     }
     break;
   case dwarf::DW_ATE_signed_char:
     if (ByteSize == 1)
       STK = SimpleTypeKind::SignedCharacter;
     break;

llvm/lib/DebugInfo/CodeView/TypeIndex.cpp

static const SimpleTypeEntry SimpleTypeNames[] = {
     {"<not translated>*", SimpleTypeKind::NotTranslated},
     {"HRESULT*", SimpleTypeKind::HResult},
     {"signed char*", SimpleTypeKind::SignedCharacter},
     {"unsigned char*", SimpleTypeKind::UnsignedCharacter},
     {"char*", SimpleTypeKind::NarrowCharacter},
     {"wchar_t*", SimpleTypeKind::WideCharacter},
     {"char16_t*", SimpleTypeKind::Character16},
     {"char32_t*", SimpleTypeKind::Character32},
+    {"char8_t*", SimpleTypeKind::Character8},
     {"__int8*", SimpleTypeKind::SByte},
     {"unsigned __int8*", SimpleTypeKind::Byte},
     {"short*", SimpleTypeKind::Int16Short},
     {"unsigned short*", SimpleTypeKind::UInt16Short},
     {"__int16*", SimpleTypeKind::Int16},
     {"unsigned __int16*", SimpleTypeKind::UInt16},
     {"long*", SimpleTypeKind::Int32Long},
     {"unsigned long*", SimpleTypeKind::UInt32Long},

llvm/lib/DebugInfo/PDB/Native/NativeTypeEnum.cpp

PDB_BuiltinType NativeTypeEnum::getBuiltinType() const {
   case SimpleTypeKind::SignedCharacter:
     return PDB_BuiltinType::Char;
   case SimpleTypeKind::WideCharacter:
     return PDB_BuiltinType::WCharT;
   case SimpleTypeKind::Character16:
     return PDB_BuiltinType::Char16;
   case SimpleTypeKind::Character32:
     return PDB_BuiltinType::Char32;
+  case SimpleTypeKind::Character8:
+    return PDB_BuiltinType::Char8;
   case SimpleTypeKind::Int128:
   case SimpleTypeKind::Int128Oct:
   case SimpleTypeKind::Int16:
   case SimpleTypeKind::Int16Short:
   case SimpleTypeKind::Int32:
   case SimpleTypeKind::Int32Long:
   case SimpleTypeKind::Int64:
   case SimpleTypeKind::Int64Quad:

llvm/lib/DebugInfo/PDB/Native/SymbolCache.cpp

uint32_t Size;
     {codeview::SimpleTypeKind::Int32Long, PDB_BuiltinType::Int, 4},
     {codeview::SimpleTypeKind::UInt32Long, PDB_BuiltinType::UInt, 4},
     {codeview::SimpleTypeKind::Int64Quad, PDB_BuiltinType::Int, 8},
     {codeview::SimpleTypeKind::UInt64Quad, PDB_BuiltinType::UInt, 8},
     {codeview::SimpleTypeKind::NarrowCharacter, PDB_BuiltinType::Char, 1},
     {codeview::SimpleTypeKind::WideCharacter, PDB_BuiltinType::WCharT, 2},
     {codeview::SimpleTypeKind::Character16, PDB_BuiltinType::Char16, 2},
     {codeview::SimpleTypeKind::Character32, PDB_BuiltinType::Char32, 4},
+    {codeview::SimpleTypeKind::Character8, PDB_BuiltinType::Char8, 1},
     {codeview::SimpleTypeKind::SignedCharacter, PDB_BuiltinType::Char, 1},
     {codeview::SimpleTypeKind::UnsignedCharacter, PDB_BuiltinType::UInt, 1},
     {codeview::SimpleTypeKind::Float32, PDB_BuiltinType::Float, 4},
     {codeview::SimpleTypeKind::Float64, PDB_BuiltinType::Float, 8},
     {codeview::SimpleTypeKind::Float80, PDB_BuiltinType::Float, 10},
     {codeview::SimpleTypeKind::Boolean8, PDB_BuiltinType::Bool, 1},
     // This table can be grown as necessary, but these are the only types we've
     // needed so far.