[Clang][Sema] Fix last argument not being used when comparing function template specializations when one has an explicit object argument

[MitalAshok]

Fixes #92188

[llvmbot]

@llvm/pr-subscribers-clang

Author: Mital Ashok (MitalAshok)

Changes

Fixes #92188

---

Full diff: https://github.com/llvm/llvm-project/pull/92263.diff

3 Files Affected:

• (modified) clang/docs/ReleaseNotes.rst (+2)
• (modified) clang/lib/Sema/SemaTemplateDeduction.cpp (+7-2)
• (modified) clang/test/SemaCXX/cxx2b-deducing-this.cpp (+49)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index ae699ebfc6038..9850fc9f8e0b4 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -713,6 +713,8 @@ Bug Fixes to C++ Support
 - Correctly treat the compound statement of an ``if consteval`` as an immediate context. Fixes (#GH91509).
 - When partial ordering alias templates against template template parameters,
   allow pack expansions when the alias has a fixed-size parameter list. Fixes (#GH62529).
+- Fix a bug where the last argument was not considered when considering the most viable function for
+  explicit object argument member functions. Fixes (#GH92188).
 
 Bug Fixes to AST Handling
 ^^^^^^^^^^^^^^^^^^^^^^^^^
diff --git a/clang/lib/Sema/SemaTemplateDeduction.cpp b/clang/lib/Sema/SemaTemplateDeduction.cpp
index 853c0e1b50619..094f767399e2a 100644
--- a/clang/lib/Sema/SemaTemplateDeduction.cpp
+++ b/clang/lib/Sema/SemaTemplateDeduction.cpp
@@ -5522,7 +5522,8 @@ static bool isAtLeastAsSpecializedAs(Sema &S, SourceLocation Loc,
 /// function templates.
 ///
 /// \param NumCallArguments1 The number of arguments in the call to FT1, used
-/// only when \c TPOC is \c TPOC_Call.
+/// only when \c TPOC is \c TPOC_Call. Does not include the object argument when
+/// calling a member function.
 ///
 /// \param RawObj1Ty The type of the object parameter of FT1 if a member
 /// function only used if \c TPOC is \c TPOC_Call and FT1 is a Function
@@ -5591,7 +5592,11 @@ FunctionTemplateDecl *Sema::getMoreSpecializedTemplate(
                                               IsRValRef1);
       Args2.push_back(Obj2Ty);
     }
-    size_t NumComparedArguments = NumCallArguments1 + ShouldConvert1;
+    size_t NumComparedArguments = NumCallArguments1;
+    // Either added an argument above or the prototype includes an explicit
+    // object argument we need to count
+    if (Method1)
+      ++NumComparedArguments;
 
     Args1.insert(Args1.end(), Proto1->param_type_begin(),
                  Proto1->param_type_end());
diff --git a/clang/test/SemaCXX/cxx2b-deducing-this.cpp b/clang/test/SemaCXX/cxx2b-deducing-this.cpp
index 5f29a955e053c..55792e1448ac5 100644
--- a/clang/test/SemaCXX/cxx2b-deducing-this.cpp
+++ b/clang/test/SemaCXX/cxx2b-deducing-this.cpp
@@ -838,3 +838,52 @@ int h() {
   }();
 }
 }
+
+namespace GH92188 {
+struct A {
+  template<auto N>
+  void operator+=(this auto &&, const char (&)[N]);
+  void operator+=(this auto &&, auto &&) = delete;
+
+  void f1(this A &, auto &);
+  void f1(this A &, auto &&) = delete;
+
+  void f2(this auto&);
+  void f2(this auto&&) = delete;
+
+  void f3(auto&) &;
+  void f3(this A&, auto&&) = delete;
+
+  void f4(auto&&) & = delete;
+  void f4(this A&, auto&);
+
+  static void f5(auto&);
+  void f5(this A&, auto&&) = delete;
+
+  static void f6(auto&&) = delete;
+  void f6(this A&, auto&);
+
+  void implicit_this() {
+    int lval;
+    operator+=("123");
+    f1(lval);
+    f2();
+    f3(lval);
+    f4(lval);
+    f5(lval);
+    f6(lval);
+  }
+};
+
+void g() {
+  A a;
+  int lval;
+  a += "123";
+  a.f1(lval);
+  a.f2();
+  a.f3(lval);
+  a.f4(lval);
+  a.f5(lval);
+  a.f6(lval);
+}
+}

[MitalAshok]

See also: #83279

[Sirraide]

Fix seems reasonable to me.

[mizvekov]

LGTM, thanks for the fix!

[Sirraide]

@MitalAshok Just a reminder to resolve the merge conflicts in the release notes locally so we can get this merged. (I’d normally just do that myself when merging because of how common merge conflicts in the release notes are, but Github says it’s ‘too complex for the web editor’, apparently...)

[MitalAshok]

@Sirraide Oh I didn't see the conflict, thanks. It should be fine now?

[Sirraide]

Oh I didn't see the conflict

It may not have been there before; conflicts in the release notes happen all the time