[llvm] [llvm][DebugInfo] Attach object-pointer to DISubprogram declarations (PR #122742)

[Michael Buch via llvm-commits]

https://github.com/Michael137 created https://github.com/llvm/llvm-project/pull/122742

Currently Clang only attaches `DW_AT_object_pointer` to `DW_TAG_subprogram` definitions. LLDB constructs C++ method types from their `DW_TAG_subprogram` declaration, which is also the point at which it needs to determine whether a method is static or not. LLDB's heuristic for this could be very simple if we emitted `DW_AT_object_pointer` on declarations. But since we don't, LLDB has to guess whether an argument is an implicit object parameter based on the DW_AT_name and DW_AT_type.

To simplify LLDB's job (and to eventually support C++23's explicit object parameters), this patch adds the `DIFlagObjectPointer` to `DISubprogram` declarations.

For reference, GCC attaches the object-pointer DIE to both the definition and declaration: https://godbolt.org/z/3TWjTfWon

Fixes https://github.com/llvm/llvm-project/issues/120973

>From c1e4faf70993f59734882d9a98024d44eb50d82d Mon Sep 17 00:00:00 2001
From: Michael Buch <michaelbuch12 at gmail.com>
Date: Thu, 9 Jan 2025 14:37:41 +0000
Subject: [PATCH] [llvm][DebugInfo] Attach object-pointer to DISubprogram
 declarations

Currently Clang only attaches `DW_AT_object_pointer` to
`DW_TAG_subprogram` definitions. LLDB constructs C++ method
types from their `DW_TAG_subprogram` declaration, which is also
the point at which it needs to determine whether a method is
static or not. LLDB's heuristic for this could be very simple if we
emitted `DW_AT_object_pointer` on declarations. But since we don't,
LLDB has to guess whether an argument is an implicit object parameter
based on the DW_AT_name and DW_AT_type.

To simplify LLDB's job (and to eventually support C++23's explicit
object parameters), this patch adds the `DIFlagObjectPointer` to
`DISubprogram` declarations.

For reference, GCC attaches the object-pointer DIE to both the
definition and declaration: https://godbolt.org/z/3TWjTfWon

Fixes https://github.com/llvm/llvm-project/issues/120973
---
 llvm/lib/CodeGen/AsmPrinter/DwarfUnit.cpp       | 13 +++++++++++--
 llvm/lib/CodeGen/AsmPrinter/DwarfUnit.h         |  4 +++-
 llvm/test/DebugInfo/X86/DW_AT_object_pointer.ll | 10 +++++++++-
 llvm/test/DebugInfo/X86/dwarf-public-names.ll   |  2 +-
 4 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.cpp b/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.cpp
index 0a8a1ad38c959f2..d3450b8b0556fde 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.cpp
@@ -849,7 +849,9 @@ void DwarfUnit::constructTypeDIE(DIE &Buffer, const DIDerivedType *DTy) {
   }
 }
 
-void DwarfUnit::constructSubprogramArguments(DIE &Buffer, DITypeRefArray Args) {
+DIE *DwarfUnit::constructSubprogramArguments(DIE &Buffer, DITypeRefArray Args) {
+  // Args[0] is the return type.
+  DIE *ObjectPointer = nullptr;
   for (unsigned i = 1, N = Args.size(); i < N; ++i) {
     const DIType *Ty = Args[i];
     if (!Ty) {
@@ -860,8 +862,14 @@ void DwarfUnit::constructSubprogramArguments(DIE &Buffer, DITypeRefArray Args) {
       addType(Arg, Ty);
       if (Ty->isArtificial())
         addFlag(Arg, dwarf::DW_AT_artificial);
+      if (Ty->isObjectPointer()) {
+        assert(!ObjectPointer && "Can't have more than one object pointer");
+        ObjectPointer = &Arg;
+      }
     }
   }
+
+  return ObjectPointer;
 }
 
 void DwarfUnit::constructTypeDIE(DIE &Buffer, const DISubroutineType *CTy) {
@@ -1358,7 +1366,8 @@ void DwarfUnit::applySubprogramAttributes(const DISubprogram *SP, DIE &SPDie,
 
     // Add arguments. Do not add arguments for subprogram definition. They will
     // be handled while processing variables.
-    constructSubprogramArguments(SPDie, Args);
+    if (auto *ObjectPointer = constructSubprogramArguments(SPDie, Args))
+      addDIEEntry(SPDie, dwarf::DW_AT_object_pointer, *ObjectPointer);
   }
 
   addThrownTypes(SPDie, SP->getThrownTypes());
diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.h b/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.h
index 163205378fb4b66..7a5295d826a483b 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.h
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfUnit.h
@@ -268,7 +268,9 @@ class DwarfUnit : public DIEUnit {
   void constructContainingTypeDIEs();
 
   /// Construct function argument DIEs.
-  void constructSubprogramArguments(DIE &Buffer, DITypeRefArray Args);
+  ///
+  /// \returns DIE of the object pointer if one exists. Nullptr otherwise.
+  DIE *constructSubprogramArguments(DIE &Buffer, DITypeRefArray Args);
 
   /// Create a DIE with the given Tag, add the DIE to its parent, and
   /// call insertDIE if MD is not null.
diff --git a/llvm/test/DebugInfo/X86/DW_AT_object_pointer.ll b/llvm/test/DebugInfo/X86/DW_AT_object_pointer.ll
index d9988ac31451e2c..30d4203466766ef 100644
--- a/llvm/test/DebugInfo/X86/DW_AT_object_pointer.ll
+++ b/llvm/test/DebugInfo/X86/DW_AT_object_pointer.ll
@@ -5,7 +5,15 @@
 ; CHECK-NOT: ""
 ; CHECK: DW_TAG
 ; CHECK: DW_TAG_class_type
-; CHECK: DW_AT_object_pointer [DW_FORM_ref4]     (cu + 0x{{[0-9a-f]*}} => {[[PARAM:0x[0-9a-f]*]]})
+; CHECK: [[DECL:0x[0-9a-f]+]]: DW_TAG_subprogram
+; CHECK:                         DW_AT_name {{.*}} "A"
+; CHECK:                         DW_AT_object_pointer [DW_FORM_ref4] 
+; CHECK-SAME:                    (cu + 0x{{[0-9a-f]*}} => {[[DECL_PARAM:0x[0-9a-f]*]]})
+; CHECK: [[DECL_PARAM]]:         DW_TAG_formal_parameter
+;
+; CHECK: DW_TAG_subprogram
+; CHECK:   DW_AT_specification [DW_FORM_ref4] (cu + {{.*}} => {[[DECL]]}
+; CHECK:   DW_AT_object_pointer [DW_FORM_ref4]     (cu + 0x{{[0-9a-f]*}} => {[[PARAM:0x[0-9a-f]*]]})
 ; CHECK: [[PARAM]]:     DW_TAG_formal_parameter
 ; CHECK-NOT: DW_TAG
 ; CHECK: DW_AT_name [DW_FORM_strp]     ( .debug_str[0x{{[0-9a-f]*}}] = "this")
diff --git a/llvm/test/DebugInfo/X86/dwarf-public-names.ll b/llvm/test/DebugInfo/X86/dwarf-public-names.ll
index c2274511d4191f4..a484c094892d0cf 100644
--- a/llvm/test/DebugInfo/X86/dwarf-public-names.ll
+++ b/llvm/test/DebugInfo/X86/dwarf-public-names.ll
@@ -61,7 +61,7 @@
 
 ; Skip the output to the header of the pubnames section.
 ; LINUX: debug_pubnames
-; LINUX-NEXT: unit_size = 0x00000128
+; LINUX-NEXT: unit_size =
 
 ; Check for each name in the output.
 ; LINUX-DAG: "ns"