[CIR] Treat cir.record class and struct types equivalently. (#1564)

When we create a `cir.record` type, we track the `RecordKind` according
to whether the source code declared the type using `struct` or `class`
even though these are semantically equivalent. The distinction is used
in naming the type when we lower to the LLVM dialect.

This change updates the code to remove the last place where we were
handling class and struct records differently.

Author: andykaylor

clang/lib/CIR/Dialect/IR/CIRDialect.cpp

@@ -3613,10 +3613,7 @@ LogicalResult cir::GetMemberOp::verify() {
   if (recordTy.getMembers().size() <= getIndex())
     return emitError() << "member index out of bounds";
 
-  // FIXME(cir): member type check is disabled for classes as the codegen for
-  // these still need to be patched.
-  if (!recordTy.isClass() &&
-      recordTy.getMembers()[getIndex()] != getResultTy().getPointee())
+  if (recordTy.getMembers()[getIndex()] != getResultTy().getPointee())
     return emitError() << "member type mismatch";
 
   return mlir::success();

clang/test/CIR/IR/getmember.cir

@@ -4,8 +4,6 @@
 !u16i = !cir.int<u, 16>
 !u32i = !cir.int<u, 32>
 
-!ty_Class = !cir.record<class "Class" {!u16i, !u32i}>
-!ty_Incomplete = !cir.record<struct "Incomplete" incomplete>
 !ty_Struct = !cir.record<struct "Struct" {!u16i, !u32i}>
 
 module  {
@@ -14,11 +12,4 @@ module  {
     %0 = cir.get_member %arg0[1] {name = "test"} : !cir.ptr<!ty_Struct> -> !cir.ptr<!u32i>
     cir.return
   }
-
-  // FIXME: remove bypass once codegen for CIR class records is patched.
-  cir.func @shouldBypassMemberTypeCheckForClassRecords(%arg0 : !cir.ptr<!ty_Class>) {
-    // CHECK: cir.get_member %arg0[1] {name = "test"} : !cir.ptr<!ty_Class> -> !cir.ptr<!cir.ptr<!u32i>> 
-    %0 = cir.get_member %arg0[1] {name = "test"} : !cir.ptr<!ty_Class> -> !cir.ptr<!cir.ptr<!u32i>>
-    cir.return
-  }
 }