[clang] [clang-tools-extra] [RecursiveASTVisitor] Skip implicit instantiations. (PR #110899)

[Harald van Dijk via cfe-commits]

================
@@ -2069,22 +2069,24 @@ bool RecursiveASTVisitor<Derived>::TraverseTemplateArgumentLocsHelper(
 
 #define DEF_TRAVERSE_TMPL_SPEC_DECL(TMPLDECLKIND, DECLKIND)                    \
   DEF_TRAVERSE_DECL(TMPLDECLKIND##TemplateSpecializationDecl, {                \
+    auto TSK = D->getTemplateSpecializationKind();                             \
     /* For implicit instantiations ("set<int> x;"), we don't want to           \
        recurse at all, since the instatiated template isn't written in         \
        the source code anywhere.  (Note the instatiated *type* --              \
        set<int> -- is written, and will still get a callback of                \
        TemplateSpecializationType).  For explicit instantiations               \
        ("template set<int>;"), we do need a callback, since this               \
-       is the only callback that's made for this instantiation.                \
-       We use getTemplateArgsAsWritten() to distinguish. */                    \
-    if (const auto *ArgsWritten = D->getTemplateArgsAsWritten()) {             \
-      /* The args that remains unspecialized. */                               \
-      TRY_TO(TraverseTemplateArgumentLocsHelper(                               \
-          ArgsWritten->getTemplateArgs(), ArgsWritten->NumTemplateArgs));      \
+       is the only callback that's made for this instantiation. */             \
+    if (TSK != TSK_ImplicitInstantiation) {                                    \
----------------
hvdijk wrote:

Thanks. It doesn't seem like the same situation as `TraverseFunctionHelper` to me: there, the logic is that if it's an implicit instantiation, that instantiation isn't written in the source code anywhere, so should be skipped by AST traversal. But the check in `TraverseFunctionHelper` assumes we possibly did end up in an implicit instantiation anyway, and then it needs to be handled in some appropriate way. Still, I think you're probably right that this should check `TSK_Undeclared` as well. I don't think that can be covered by this clang-tidy check, but I can at least check that it doesn't break any other tests.

https://github.com/llvm/llvm-project/pull/110899