[TPDE][LLVM] Add support for debug line info

tpde-llvm/src/LLVMCompilerBase.hpp

@@ -25,6 +25,7 @@
 
 #include "tpde/Assembler.hpp"
 #include "tpde/CompilerBase.hpp"
+#include "tpde/DwarfLineWriter.hpp"
 #include "tpde/ValLocalIdx.hpp"
 #include "tpde/ValueAssignment.hpp"
 #include "tpde/base.hpp"
@@ -169,6 +170,8 @@ struct LLVMCompilerBase : public LLVMCompiler,
 
   tpde::util::SmallVector<std::pair<IRValueRef, SymRef>, 16> type_info_syms;
 
+  std::unique_ptr<tpde::DwarfLineWriter> dw_line_writer;
+
   enum class LibFunc {
     divti3,
     udivti3,
@@ -335,6 +338,66 @@ struct LLVMCompilerBase : public LLVMCompiler,
   }
 
 public:
+  /// Record a debug location at the current PC
+  void record_debug_location(llvm::DebugLoc debug_loc) {
+    if (!dw_line_writer || !debug_loc) {
+      return;
+    }
+
+    const u32 current_pc = this->text_writer.offset();
+    // Avoid emitting duplicate locations for the same PC, which can happen
+    // when multiple instructions share the same debug location.
+    if (dw_line_writer->offset_equals_last_offset(current_pc)) {
+      return;
+    }
+
+    const llvm::DILocation *di_loc = debug_loc.get();
+    const u32 line = di_loc->getLine();
+    // Line number 0 is used to indicate an unknown location, so skip it.
+    if (line == 0) {
+      return;
+    }
+
+    dw_line_writer->push_location(
+        current_pc,
+        tpde::DebugLocation{
+            .file_name = di_loc->getFilename(),
+            .directory = di_loc->getDirectory(),
+            .line = di_loc->getLine(),
+            .column = static_cast<u16>(di_loc->getColumn()),
+        });
+  }
+
+  void set_current_debug_location_function(const IRFuncRef func) {
+    if (!dw_line_writer) {
+      return;
+    }
+
+    const u32 current_pc = this->text_writer.offset();
+    if (const llvm::DISubprogram *sp = func->getSubprogram()) {
+      dw_line_writer->push_location(
+        current_pc,
+        tpde::DebugLocation{
+            .file_name = sp->getFilename(),
+            .directory = sp->getDirectory(),
+            .line = sp->getLine(),
+            .column = 0u,
+        });
+      dw_line_writer->set_subprogram_written();
+    }
+  }
+
+  void finish_debug_location_function(const IRFuncRef, const u32 func_idx) {
+    if (!dw_line_writer) {
+      return;
+    }
+
+    const u32 sym_id = this->func_syms[func_idx].id();
+    const auto it = this->func_sym_id_to_skew.find(sym_id);
+    const u32 skew = (it != this->func_sym_id_to_skew.end()) ? it->second : 0u;
+    dw_line_writer->end_function(skew);
+  }
+
   /// Whether to use a DSO-local access instead of going through the GOT.
   static bool use_local_access(const llvm::GlobalValue *gv) {
     // If the symbol is preemptible, don't generate a local access.
@@ -1250,6 +1313,11 @@ bool LLVMCompilerBase<Adaptor, Derived, Config>::compile(llvm::Module &mod) {
   group_secs.clear();
   libfunc_syms.fill({});
 
+  if (mod.getNamedMetadata("llvm.dbg.cu") != nullptr) {
+    const tpde::DwarfConfig cfg(Config::MIN_INST_WIDTH);
+    dw_line_writer = std::make_unique<tpde::DwarfLineWriter>(cfg);
+  }
+
   if (!Base::compile()) {
     return false;
   }
@@ -1270,6 +1338,12 @@ bool LLVMCompilerBase<Adaptor, Derived, Config>::compile(llvm::Module &mod) {
     this->assembler.sym_copy(dst_sym, from_sym);
   }
 
+  // Write DWARF 5 line information if enabled and the module has debug info.
+  if (dw_line_writer) {
+    llvm::TimeTraceScope time_scope("TPDE_DebugLine_Write");
+    dw_line_writer->finalize(this->assembler);
+  }
+
   return true;
 }
 
@@ -1366,6 +1440,9 @@ bool LLVMCompilerBase<Adaptor, Derived, Config>::compile_inst(
     return res;
   }();
 
+  // Record debug location before instruction
+  record_debug_location(i->getDebugLoc());
+
   const ValInfo &val_info = this->adaptor->val_info(i);
   assert(i->getOpcode() < fns.size());
   const auto [compile_fn, arg] = fns[i->getOpcode()];

tpde/CMakeLists.txt

@@ -33,6 +33,7 @@ target_sources(tpde PRIVATE
     src/Assembler.cpp
     src/AssemblerElf.cpp
     src/base.cpp
+    src/DwarfLineWriter.cpp
     src/ElfMapper.cpp
     src/FunctionWriter.cpp
     src/StringTable.cpp
@@ -52,6 +53,7 @@ target_sources(tpde PRIVATE
         include/tpde/CompilerBase.hpp
         include/tpde/CompilerConfig.hpp
         include/tpde/DWARF.hpp
+        include/tpde/DwarfLineWriter.hpp
         include/tpde/ELF.hpp
         include/tpde/ElfMapper.hpp
         include/tpde/FunctionWriter.hpp

tpde/include/tpde/Assembler.hpp

@@ -53,15 +53,18 @@ struct Relocation {
 
 /// Section kinds, lowered to file-format specific flags.
 enum class SectionKind : u8 {
-  Text,       ///< Text section, executable code (ELF .text)
-  ReadOnly,   ///< Read-only data section (ELF .rodata)
-  EHFrame,    ///< EH Frame section (ELF .eh_frame)
-  LSDA,       ///< LSDA section (ELF .gcc_except_table)
-  Data,       ///< Writable data section (ELF .data)
-  DataRelRO,  ///< Read-only data section with relocations (ELF .data.rel.ro)
-  BSS,        ///< Zero-initialized data section (ELF .bss)
-  ThreadData, ///< Initialized thread-local data section (ELF .tdata)
-  ThreadBSS,  ///< Zero-initialized thread-local data section (ELF .tbss)
+  Text,        ///< Text section, executable code (ELF .text)
+  ReadOnly,    ///< Read-only data section (ELF .rodata)
+  EHFrame,     ///< EH Frame section (ELF .eh_frame)
+  LSDA,        ///< LSDA section (ELF .gcc_except_table)
+  Data,        ///< Writable data section (ELF .data)
+  DataRelRO,   ///< Read-only data section with relocations (ELF .data.rel.ro)
+  BSS,         ///< Zero-initialized data section (ELF .bss)
+  ThreadData,  ///< Initialized thread-local data section (ELF .tdata)
+  ThreadBSS,   ///< Zero-initialized thread-local data section (ELF .tbss)
+  DebugInfo,   ///< DWARF debug info (ELF .debug_info)
+  DebugLine,   ///< DWARF line information (ELF .debug_line)
+  DebugAbbrev, ///< DWARF abbreviations (ELF .debug_abbrev)
 
   Max
 };

tpde/include/tpde/CompilerBase.hpp

@@ -194,6 +194,8 @@ struct CompilerBase {
   // allocations
   util::SmallVector<Label> block_labels;
 
+  std::unordered_map<u32, u32> func_sym_id_to_skew; // Map SymRef.id() to address skew
+
   util::SmallVector<std::pair<SymRef, SymRef>, 4> personality_syms = {};
 
   struct ScratchReg;
@@ -578,6 +580,10 @@ struct CompilerBase {
 
   void analysis_end() {}
 
+  void set_current_debug_location_function(const IRFuncRef) {}
+
+  void finish_debug_location_function(const IRFuncRef, const u32) {}
+
   void reloc_text(SymRef sym, u32 type, u64 offset, i64 addend = 0) {
     this->assembler.reloc_sec(
         text_writer.get_sec_ref(), sym, type, offset, addend);
@@ -2315,6 +2321,8 @@ bool CompilerBase<Adaptor, Derived, Config>::compile_func(const IRFuncRef func,
 
   derived()->start_func(func_idx);
 
+  derived()->set_current_debug_location_function(func);
+
   block_labels.clear();
   block_labels.resize_uninitialized(analyzer.block_layout.size());
   for (u32 i = 0; i < analyzer.block_layout.size(); ++i) {
@@ -2403,6 +2411,7 @@ bool CompilerBase<Adaptor, Derived, Config>::compile_func(const IRFuncRef func,
          "found non-freed ValueAssignment, maybe missing ref-count?");
 
   derived()->finish_func(func_idx);
+  derived()->finish_debug_location_function(func, func_idx);
   this->text_writer.finish_func();
 
   return true;

tpde/include/tpde/CompilerConfig.hpp

@@ -22,6 +22,7 @@ concept CompilerConfig = requires {
   { T::PLATFORM_POINTER_SIZE } -> SameBaseAs<u32>;
   { T::NUM_BANKS } -> SameBaseAs<u32>;
   { T::DEFAULT_VAR_REF_HANDLING } -> SameBaseAs<bool>;
+  { T::MIN_INST_WIDTH } -> SameBaseAs<u8>;
 
   typename T::DefaultCCAssigner;
   requires std::derived_from<typename T::DefaultCCAssigner, CCAssigner>;
@@ -33,6 +34,7 @@ concept CompilerConfig = requires {
 
 struct CompilerConfigDefault {
   constexpr static bool DEFAULT_VAR_REF_HANDLING = true;
+  constexpr static u8 MIN_INST_WIDTH = 1;
 };
 
 } // namespace tpde

tpde/include/tpde/DWARF.hpp

@@ -34,6 +34,45 @@ constexpr u8 DWARF_CFI_PRIMARY_OPCODE_MASK = 0xc0;
 
 constexpr u32 EH_FDE_FUNC_START_OFF = 0x8;
 
+constexpr u8 DW_LNS_extended_op = 0;
+constexpr u8 DW_LNS_copy = 1;
+constexpr u8 DW_LNS_advance_pc = 2;
+constexpr u8 DW_LNS_advance_line = 3;
+constexpr u8 DW_LNS_set_file = 4;
+constexpr u8 DW_LNS_set_column = 5;
+constexpr u8 DW_LNS_negate_stmt = 6;
+constexpr u8 DW_LNS_set_basic_block = 7;
+constexpr u8 DW_LNS_const_add_pc = 8;
+constexpr u8 DW_LNS_fixed_advance_pc = 9;
+constexpr u8 DW_LNS_set_prologue_end = 10;
+constexpr u8 DW_LNS_set_epilogue_begin = 11;
+constexpr u8 DW_LNS_set_isa = 12;
+
+constexpr u8 DW_LNE_end_sequence = 1;
+constexpr u8 DW_LNE_set_address = 2;
+constexpr u8 DW_LNE_define_file = 3;
+constexpr u8 DW_LNE_set_discriminator = 4;
+
+constexpr u8 DW_TAG_compile_unit = 17;
+
+constexpr u8 DW_AT_stmt_list = 16;
+constexpr u8 DW_AT_low_pc = 17;
+constexpr u8 DW_AT_high_pc = 18;
+
+constexpr u8 DW_FORM_addr = 1;
+constexpr u8 DW_FORM_string = 8;
+constexpr u8 DW_FORM_udata = 0x0f;
+constexpr u8 DW_FORM_sec_offset = 23;
+
+// DWARF 5 line content type codes
+constexpr u8 DW_LNCT_path = 1;
+constexpr u8 DW_LNCT_directory_index = 2;
+
+// DWARF 5 unit type codes
+constexpr u8 DW_UT_compile = 0x01;
+
+constexpr u8 DW_CHILDREN_no = 0;
+
 namespace x64 {
 constexpr u8 DW_reg_rax = 0;
 constexpr u8 DW_reg_rdx = 1;