[clang] [CIR] Upstream AssignmentMemcpyizer (PR #236803)

Steffen Larsen via cfe-commits cfe-commits at lists.llvm.org
Thu Nov 5 00:33:18 PST 2026


================
@@ -873,17 +1140,10 @@ void CIRGenFunction::emitImplicitAssignmentOperatorBody(FunctionArgList &args) {
   assert(!cir::MissingFeatures::incrementProfileCounter());
   assert(!cir::MissingFeatures::runCleanupsScope());
 
-  // Classic codegen uses a special class to attempt to replace member
-  // initializers with memcpy. We could possibly defer that to the
-  // lowering or optimization phases to keep the memory accesses more
-  // explicit. For now, we don't insert memcpy at all, though in some
-  // cases the AST contains a call to memcpy.
-  assert(!cir::MissingFeatures::assignMemcpyizer());
-  for (Stmt *s : rootCS->body())
-    if (emitStmt(s, /*useCurrentScope=*/true).failed())
-      cgm.errorNYI(s->getSourceRange(),
-                   std::string("emitImplicitAssignmentOperatorBody: ") +
-                       s->getStmtClassName());
+  AssignmentMemcpyizer AM(*this, assignOp, args);
+  for (auto *i : rootCS->body())
+    AM.emitAssignment(i);
+  AM.finish();
----------------
steffenlarsen wrote:

Do we think another kind of pattern will ever exist for the usage of this? Otherwise, we could move this logic to a public static member of `AssignmentMemcpyizer` and make all the other members private. I think my main worries are:
1. Unintentionally dropping aggregated statements due to a missing flush.
2. Someone inadverdently adding code before the finish that causes invalid order of emission.

https://github.com/llvm/llvm-project/pull/236803


More information about the cfe-commits mailing list