llvm-dwarfinfo add support to decode DW_OP_convert

[markus]

Summary

Added llvm-dwarfdump support to decode DW_OP_convert (and eventually other typed DW5 ops) as a preparation before introducing such ops in llvm debug info output (see https://reviews.llvm.org/D56587)

Example output

0x0000013d:     DW_TAG_variable
                  DW_AT_location        (0x000000d3
                     [0x0000000000401197,  0x000000000040119a): DW_OP_breg0 RAX+0, DW_OP_constu 0xff, DW_OP_and, DW_OP_convert (0x00000107) "DW_ATE_signed_8", DW_OP_convert (0x0000010e) "DW_ATE_signed_16", DW_OP_stack_value)
                  DW_AT_name    ("z")
                  DW_AT_decl_file       ("/repo/elavkje/llvm/my-dbg-test/dwarf-sext/linked/dbg-bar.c")
                  DW_AT_decl_line       (3)
                  DW_AT_type    (0x0000014d "short")

0x0000014c:     NULL

[aprantl]

That's awesome!

test/tools/llvm-dwarfdump/X86/debug_expr_convert.s

2

Can you change this or add a second invocation of RUN: not llvm-dwarfdump -verify %t.o?

[markus]

Added support for rudimentary verify of expressions (basically to check that a BaseTypeRef operand actually points to a DW_TAG_base_type DIE).

[aprantl]

lib/DebugInfo/DWARF/DWARFExpression.cpp

304

It would be nice to bubble up a more specific error message.

include/llvm/DebugInfo/DWARF/DWARFDebugLoc.h

public:
   struct LocationList {
     /// The beginning offset where this location list is stored in the debug_loc
     /// section.
     unsigned Offset;
     /// All the locations in which the variable is stored.
     SmallVector<Entry, 2> Entries;
     /// Dump this list on OS.
     void dump(raw_ostream &OS, bool IsLittleEndian, unsigned AddressSize,
-              const MCRegisterInfo *MRI, uint64_t BaseAddress,
+              const MCRegisterInfo *MRI, DWARFUnit *U, uint64_t BaseAddress,
               unsigned Indent) const;
   };
 
 private:
   using LocationLists = SmallVector<LocationList, 4>;
 
   /// A list of all the variables in the debug_loc section, each one describing
   /// the locations in which the variable is stored.

include/llvm/DebugInfo/DWARF/DWARFExpression.h

enum Encoding : uint8_t {
       Size1 = 0,
       Size2 = 1,
       Size4 = 2,
       Size8 = 3,
       SizeLEB = 4,
       SizeAddr = 5,
       SizeRefAddr = 6,
       SizeBlock = 7, ///< Preceding operand contains block size
-      SignBit = 0x8,
+      BaseTypeRef = 8,
+      SignBit = 0x80,
       SignedSize1 = SignBit | Size1,
       SignedSize2 = SignBit | Size2,
       SignedSize4 = SignBit | Size4,
       SignedSize8 = SignBit | Size8,
       SignedSizeLEB = SignBit | SizeLEB,
       SizeNA = 0xFF ///< Unused operands get this encoding.
     };
 
     enum DwarfVersion : uint8_t {
       DwarfNA, ///< Serves as a marker for unused entries
       Dwarf2 = 2,
       Dwarf3,
-      Dwarf4
+      Dwarf4,
+      Dwarf5
     };
 
     /// Description of the encoding of one expression Op.
     struct Description {
       DwarfVersion Version; ///< Dwarf version where the Op was introduced.
       Encoding Op[2];       ///< Encoding for Op operands, or SizeNA.
 
       Description(DwarfVersion Version = DwarfNA, Encoding Op1 = SizeNA,
public:
   public:
     Description &getDescription() { return Desc; }
     uint8_t getCode() { return Opcode; }
     uint64_t getRawOperand(unsigned Idx) { return Operands[Idx]; }
     uint32_t getEndOffset() { return EndOffset; }
     bool extract(DataExtractor Data, uint16_t Version, uint8_t AddressSize,
                  uint32_t Offset);
     bool isError() { return Error; }
-    bool print(raw_ostream &OS, const DWARFExpression *U,
-               const MCRegisterInfo *RegInfo, bool isEH);
+    bool print(raw_ostream &OS, const DWARFExpression *Expr,
+               const MCRegisterInfo *RegInfo, DWARFUnit *U, bool isEH);
   };
 
   /// An iterator to go through the expression operations.
   class iterator
       : public iterator_facade_base<iterator, std::forward_iterator_tag,
                                     Operation> {
     friend class DWARFExpression;
     const DWARFExpression *Expr;
public:
   DWARFExpression(DataExtractor Data, uint16_t Version, uint8_t AddressSize)
       : Data(Data), Version(Version), AddressSize(AddressSize) {
     assert(AddressSize == 8 || AddressSize == 4);
   }
 
   iterator begin() const { return iterator(this, 0); }
   iterator end() const { return iterator(this, Data.getData().size()); }
 
-  void print(raw_ostream &OS, const MCRegisterInfo *RegInfo,
+  void print(raw_ostream &OS, const MCRegisterInfo *RegInfo, DWARFUnit *U,
              bool IsEH = false) const;
 
 private:
   DataExtractor Data;
   uint16_t Version;
   uint8_t AddressSize;
 };
 

lib/DebugInfo/DWARF/DWARFDebugFrame.cpp

else
       OS << format(" %" PRId64 "*data_alignment_factor" , Operand);
     break;
   case OT_Register:
     OS << format(" reg%" PRId64, Operand);
     break;
   case OT_Expression:
     assert(Instr.Expression && "missing DWARFExpression object");
     OS << " ";
-    Instr.Expression->print(OS, MRI, IsEH);
+    Instr.Expression->print(OS, MRI, nullptr, IsEH);
     break;
   }
 }
 
 void CFIProgram::dump(raw_ostream &OS, const MCRegisterInfo *MRI, bool IsEH,
                       unsigned IndentLevel) const {
   for (const auto &Instr : Instructions) {
     uint8_t Opcode = Instr.Opcode;

lib/DebugInfo/DWARF/DWARFDebugLoc.cpp

 
 // When directly dumping the .debug_loc without a compile unit, we have to guess
 // at the DWARF version. This only affects DW_OP_call_ref, which is a rare
 // expression that LLVM doesn't produce. Guessing the wrong version means we
 // won't be able to pretty print expressions in DWARF2 binaries produced by
 // non-LLVM tools.
 static void dumpExpression(raw_ostream &OS, ArrayRef<char> Data,
                            bool IsLittleEndian, unsigned AddressSize,
-                           const MCRegisterInfo *MRI) {
+                           const MCRegisterInfo *MRI, DWARFUnit *U) {
   DWARFDataExtractor Extractor(StringRef(Data.data(), Data.size()),
                                IsLittleEndian, AddressSize);
-  DWARFExpression(Extractor, dwarf::DWARF_VERSION, AddressSize).print(OS, MRI);
+  DWARFExpression(Extractor, dwarf::DWARF_VERSION, AddressSize).print(OS, MRI, U);
 }
 
 void DWARFDebugLoc::LocationList::dump(raw_ostream &OS, bool IsLittleEndian,
                                        unsigned AddressSize,
                                        const MCRegisterInfo *MRI,
+                                       DWARFUnit *U,
                                        uint64_t BaseAddress,
                                        unsigned Indent) const {
   for (const Entry &E : Entries) {
     OS << '\n';
     OS.indent(Indent);
     OS << format("[0x%*.*" PRIx64 ", ", AddressSize * 2, AddressSize * 2,
                  BaseAddress + E.Begin);
     OS << format(" 0x%*.*" PRIx64 ")", AddressSize * 2, AddressSize * 2,
                  BaseAddress + E.End);
     OS << ": ";
 
-    dumpExpression(OS, E.Loc, IsLittleEndian, AddressSize, MRI);
+    dumpExpression(OS, E.Loc, IsLittleEndian, AddressSize, MRI, U);
   }
 }
 
 DWARFDebugLoc::LocationList const *
 DWARFDebugLoc::getLocationListAtOffset(uint64_t Offset) const {
   auto It = std::lower_bound(
       Locations.begin(), Locations.end(), Offset,
       [](const LocationList &L, uint64_t Offset) { return L.Offset < Offset; });
   if (It != Locations.end() && It->Offset == Offset)
     return &(*It);
   return nullptr;
 }
 
 void DWARFDebugLoc::dump(raw_ostream &OS, const MCRegisterInfo *MRI,
                          Optional<uint64_t> Offset) const {
   auto DumpLocationList = [&](const LocationList &L) {
     OS << format("0x%8.8x: ", L.Offset);
-    L.dump(OS, IsLittleEndian, AddressSize, MRI, 0, 12);
+    L.dump(OS, IsLittleEndian, AddressSize, MRI, nullptr, 0, 12);
     OS << "\n\n";
   };
 
   if (Offset) {
     if (auto *L = getLocationListAtOffset(*Offset))
       DumpLocationList(*L);
     return;
   }
case dwarf::DW_LLE_offset_pair:
       break;
     case dwarf::DW_LLE_base_address:
       BaseAddr = E.Value0;
       break;
     default:
       llvm_unreachable("unreachable locations list kind");
     }
 
-    dumpExpression(OS, E.Loc, IsLittleEndian, AddressSize, MRI);
+    dumpExpression(OS, E.Loc, IsLittleEndian, AddressSize, MRI, nullptr);
   }
 }
 
 void DWARFDebugLoclists::dump(raw_ostream &OS, uint64_t BaseAddr,
                               const MCRegisterInfo *MRI,
                               Optional<uint64_t> Offset) const {
   auto DumpLocationList = [&](const LocationList &L) {
     OS << format("0x%8.8x: ", L.Offset);

lib/DebugInfo/DWARF/DWARFDie.cpp

static void dumpLocation(raw_ostream &OS, DWARFFormValue &FormValue,
   const DWARFObject &Obj = Ctx.getDWARFObj();
   const MCRegisterInfo *MRI = Ctx.getRegisterInfo();
   if (FormValue.isFormClass(DWARFFormValue::FC_Block) ||
       FormValue.isFormClass(DWARFFormValue::FC_Exprloc)) {
     ArrayRef<uint8_t> Expr = *FormValue.getAsBlock();
     DataExtractor Data(StringRef((const char *)Expr.data(), Expr.size()),
                        Ctx.isLittleEndian(), 0);
     DWARFExpression(Data, U->getVersion(), U->getAddressByteSize())
-        .print(OS, MRI);
+        .print(OS, MRI, U);
     return;
   }
 
   FormValue.dump(OS, DumpOpts);
   if (FormValue.isFormClass(DWARFFormValue::FC_SectionOffset)) {
     uint32_t Offset = *FormValue.getAsSectionOffset();
     if (!U->isDWOUnit() && !U->getLocSection()->Data.empty()) {
       DWARFDebugLoc DebugLoc;
       DWARFDataExtractor Data(Obj, *U->getLocSection(), Ctx.isLittleEndian(),
                               Obj.getAddressSize());
       auto LL = DebugLoc.parseOneLocationList(Data, &Offset);
       if (LL) {
         uint64_t BaseAddr = 0;
         if (Optional<SectionedAddress> BA = U->getBaseAddress())
           BaseAddr = BA->Address;
-        LL->dump(OS, Ctx.isLittleEndian(), Obj.getAddressSize(), MRI, BaseAddr,
-                 Indent);
+        LL->dump(OS, Ctx.isLittleEndian(), Obj.getAddressSize(), MRI, U,
+                 BaseAddr, Indent);
       } else
         OS << "error extracting location list.";
       return;
     }
 
     bool UseLocLists = !U->isDWOUnit();
     StringRef LoclistsSectionData =
         UseLocLists ? Obj.getLoclistsSection().Data : U->getLocSectionData();

lib/DebugInfo/DWARF/DWARFExpression.cpp

 //===-- DWARFExpression.cpp -----------------------------------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "llvm/DebugInfo/DWARF/DWARFExpression.h"
+#include "llvm/DebugInfo/DWARF/DWARFUnit.h"
 #include "llvm/BinaryFormat/Dwarf.h"
 #include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/Support/Format.h"
 #include <cassert>
 #include <cstdint>
 #include <vector>
 
 using namespace llvm;
static DescVector getDescriptions() {
   Descriptions[DW_OP_bit_piece] = Desc(Op::Dwarf3, Op::SizeLEB, Op::SizeLEB);
   Descriptions[DW_OP_implicit_value] =
       Desc(Op::Dwarf3, Op::SizeLEB, Op::SizeBlock);
   Descriptions[DW_OP_stack_value] = Desc(Op::Dwarf3);
   Descriptions[DW_OP_GNU_push_tls_address] = Desc(Op::Dwarf3);
   Descriptions[DW_OP_addrx] = Desc(Op::Dwarf4, Op::SizeLEB);
   Descriptions[DW_OP_GNU_addr_index] = Desc(Op::Dwarf4, Op::SizeLEB);
   Descriptions[DW_OP_GNU_const_index] = Desc(Op::Dwarf4, Op::SizeLEB);
+
+  Descriptions[DW_OP_convert] = Desc(Op::Dwarf5, Op::BaseTypeRef);
+
   return Descriptions;
 }
 
 static DWARFExpression::Operation::Description getOpDesc(unsigned OpCode) {
   // FIXME: Make this constexpr once all compilers are smart enough to do it.
   static DescVector Descriptions = getDescriptions();
   // Handle possible corrupted or unsupported operation.
   if (OpCode >= Descriptions.size())
case Operation::SizeRefAddr:
       }
       break;
     case Operation::SizeLEB:
       if (Signed)
         Operands[Operand] = Data.getSLEB128(&Offset);
       else
         Operands[Operand] = Data.getULEB128(&Offset);
       break;
+    case Operation::BaseTypeRef:
+      Operands[Operand] = Data.getULEB128(&Offset);
+      break;
     case Operation::SizeBlock:
       // We need a size, so this cannot be the first operand
       if (Operand == 0)
         return false;
       // Store the offset of the block as the value.
       Operands[Operand] = Offset;
       Offset += Operands[Operand - 1];
       break;
static bool prettyPrintRegisterOp(raw_ostream &OS, uint8_t Opcode,
   }
 
   return false;
 }
 
 bool DWARFExpression::Operation::print(raw_ostream &OS,
                                        const DWARFExpression *Expr,
                                        const MCRegisterInfo *RegInfo,
+                                       DWARFUnit *U,
                                        bool isEH) {
   if (Error) {
     OS << "<decoding error>";
     return false;
   }
 
   StringRef Name = OperationEncodingString(Opcode);
   assert(!Name.empty() && "DW_OP has no name!");
   OS << Name;
 
   if ((Opcode >= DW_OP_breg0 && Opcode <= DW_OP_breg31) ||
       (Opcode >= DW_OP_reg0 && Opcode <= DW_OP_reg31) ||
       Opcode == DW_OP_bregx || Opcode == DW_OP_regx)
     if (prettyPrintRegisterOp(OS, Opcode, Operands, RegInfo, isEH))
       return true;
 
   for (unsigned Operand = 0; Operand < 2; ++Operand) {
     unsigned Size = Desc.Op[Operand];
     unsigned Signed = Size & Operation::SignBit;
 
     if (Size == Operation::SizeNA)
       break;
 
-    if (Size == Operation::SizeBlock) {
+    if (Size == Operation::BaseTypeRef && U) {
+      auto Die = U->getDIEForOffset(U->getOffset() + Operands[Operand]);
+      if (Die && Die.getTag() == dwarf::DW_TAG_base_type) {
+        OS << format(" (0x%08x)", U->getOffset() + Operands[Operand]);
+        if (auto Name = Die.find(dwarf::DW_AT_name))
+          OS << " \"" << Name->getAsCString() << "\"";
+      } else {
+        OS << format(" <invalid base_type ref: 0x%" PRIx64 ">",
+                     Operands[Operand]);
+      }
+    } else if (Size == Operation::SizeBlock) {
       uint32_t Offset = Operands[Operand];
       for (unsigned i = 0; i < Operands[Operand - 1]; ++i)
         OS << format(" 0x%02x", Expr->Data.getU8(&Offset));
     } else {
       if (Signed)
         OS << format(" %+" PRId64, (int64_t)Operands[Operand]);
       else
         OS << format(" 0x%" PRIx64, Operands[Operand]);
     }
   }
   return true;
 }
 
 void DWARFExpression::print(raw_ostream &OS, const MCRegisterInfo *RegInfo,
-                            bool IsEH) const {
+                            DWARFUnit *U, bool IsEH) const {
   for (auto &Op : *this) {
-    if (!Op.print(OS, this, RegInfo, IsEH)) {
+    if (!Op.print(OS, this, RegInfo, U, IsEH)) {
       uint32_t FailOffset = Op.getEndOffset();
       while (FailOffset < Data.getData().size())
         OS << format(" %02x", Data.getU8(&FailOffset));
       return;
     }
     if (Op.getEndOffset() < Data.getData().size())
       OS << ", ";
   }
 }
 
 } // namespace llvm

test/tools/llvm-dwarfdump/X86/debug_expr_convert.s

+# RUN: llvm-mc %s -filetype obj -triple x86_64-pc-linux -o %t.o
+# RUN: llvm-dwarfdump %t.o | FileCheck %s
+
+# CHECK: DW_AT_location  (DW_OP_addr 0x{{[0-9a-f]+}}, DW_OP_convert <invalid base_type ref: 0x{{[0-9a-f]+}}>, DW_OP_convert (0x{{[0-9a-f]+}}) "DW_ATE_signed_32")
+
+  .text
+  .file "dbg.g.c"
+  .file 1 "/tmp/dbg.g.c"
+  .type global,@object          # @global
+  .data
+  .globl  global
+  .p2align  2
+global:
+  .long 255                     # 0xff
+  .size global, 4
+
+  .section  .debug_str,"MS",@progbits,1
+.Linfo_string0:
+  .asciz  "clang version 9.0.0 (trunk 354265) (llvm/trunk 354264)" # string offset=0
+.Linfo_string1:
+  .asciz  "dbg.g.c"               # string offset=55
+.Linfo_string2:
+  .asciz  "/tmp"                  # string offset=63
+.Linfo_string3:
+  .asciz  "global"                # string offset=68
+.Linfo_string4:
+  .asciz  "int"                   # string offset=75
+.Linfo_string5:
+  .asciz  "DW_ATE_signed_32"      # string offset=79
+.Linfo_string6:
+  .asciz  "DW_ATE_signed_8"       # string offset=96
+  .section  .debug_abbrev,"",@progbits
+  .byte 1                       # Abbreviation Code
+  .byte 17                      # DW_TAG_compile_unit
+  .byte 1                       # DW_CHILDREN_yes
+  .byte 37                      # DW_AT_producer
+  .byte 14                      # DW_FORM_strp
+  .byte 19                      # DW_AT_language
+  .byte 5                       # DW_FORM_data2
+  .byte 3                       # DW_AT_name
+  .byte 14                      # DW_FORM_strp
+  .byte 16                      # DW_AT_stmt_list
+  .byte 23                      # DW_FORM_sec_offset
+  .byte 27                      # DW_AT_comp_dir
+  .byte 14                      # DW_FORM_strp
+  .byte 0                       # EOM(1)
+  .byte 0                       # EOM(2)
+  .byte 2                       # Abbreviation Code
+  .byte 36                      # DW_TAG_base_type
+  .byte 0                       # DW_CHILDREN_no
+  .byte 3                       # DW_AT_name
+  .byte 14                      # DW_FORM_strp
+  .byte 62                      # DW_AT_encoding
+  .byte 11                      # DW_FORM_data1
+  .byte 11                      # DW_AT_byte_size
+  .byte 11                      # DW_FORM_data1
+  .byte 0                       # EOM(1)
+  .byte 0                       # EOM(2)
+  .byte 3                       # Abbreviation Code
+  .byte 52                      # DW_TAG_variable
+  .byte 0                       # DW_CHILDREN_no
+  .byte 3                       # DW_AT_name
+  .byte 14                      # DW_FORM_strp
+  .byte 73                      # DW_AT_type
+  .byte 19                      # DW_FORM_ref4
+  .byte 63                      # DW_AT_external
+  .byte 25                      # DW_FORM_flag_present
+  .byte 58                      # DW_AT_decl_file
+  .byte 11                      # DW_FORM_data1
+  .byte 59                      # DW_AT_decl_line
+  .byte 11                      # DW_FORM_data1
+  .byte 2                       # DW_AT_location
+  .byte 24                      # DW_FORM_exprloc
+  .byte 0                       # EOM(1)
+  .byte 0                       # EOM(2)
+  .byte 0                       # EOM(3)
+  .section  .debug_info,"",@progbits
+.Lcu_begin0:
+  .long .Ldebug_info_end0-.Ldebug_info_start0 # Length of Unit
+.Ldebug_info_start0:
+  .short  4                       # DWARF version number
+  .long .debug_abbrev           # Offset Into Abbrev. Section
+  .byte 8                       # Address Size (in bytes)
+  .byte 1                       # Abbrev [1] 0xb:0x48 DW_TAG_compile_unit
+  .long .Linfo_string0          # DW_AT_producer
+  .short  12                      # DW_AT_language
+  .long .Linfo_string1          # DW_AT_name
+  .long .Lline_table_start0     # DW_AT_stmt_list
+  .long .Linfo_string2          # DW_AT_comp_dir
+  .byte 2                       # Abbrev [2] 0x1e:0x7 DW_TAG_base_type
+  .long .Linfo_string6          # DW_AT_name
+  .byte 5                       # DW_AT_encoding
+  .byte 1                       # DW_AT_byte_size
+  .byte 2                       # Abbrev [2] 0x25:0x7 DW_TAG_base_type
+  .long .Linfo_string5          # DW_AT_name
+  .byte 5                       # DW_AT_encoding
+  .byte 4                       # DW_AT_byte_size
+  .byte 3                       # Abbrev [3] 0x2c:0x1f DW_TAG_variable
+  .long .Linfo_string3          # DW_AT_name
+  .long 75                      # DW_AT_type
+                                        # DW_AT_external
+  .byte 1                       # DW_AT_decl_file
+  .byte 1                       # DW_AT_decl_line
+  .byte 19                      # DW_AT_location
+  .byte 3
+  .quad global
+  .byte 168
+  .asciz  "\235\200\200"  ### This offset is off by one to test the invalid base_type ref printout
+  .byte 168
+  .asciz  "\245\200\200"
+  .byte 2                       # Abbrev [2] 0x4b:0x7 DW_TAG_base_type
+  .long .Linfo_string4          # DW_AT_name
+  .byte 5                       # DW_AT_encoding
+  .byte 4                       # DW_AT_byte_size
+  .byte 0                       # End Of Children Mark
+.Ldebug_info_end0:
+  .section  .debug_macinfo,"",@progbits
+  .byte 0                       # End Of Macro List Mark
+
+  .ident  "clang version 9.0.0 (trunk 354265) (llvm/trunk 354264)"
+  .section  ".note.GNU-stack","",@progbits
+  .section  .debug_line,"",@progbits
+.Lline_table_start0: