[clang] [CIR] Upstream initial support for switch statements (PR #137106)

[Henrich Lauko via cfe-commits]

================
@@ -802,6 +804,132 @@ Block 
*cir::BrCondOp::getSuccessorForOperands(ArrayRef<Attribute> operands) {
   return nullptr;
 }
 
+//===----------------------------------------------------------------------===//
+// CaseOp
+//===----------------------------------------------------------------------===//
+
+void cir::CaseOp::getSuccessorRegions(
+    mlir::RegionBranchPoint point, SmallVectorImpl<RegionSuccessor> &regions) {
+  if (!point.isParent()) {
+    regions.push_back(RegionSuccessor());
+    return;
+  }
+  regions.push_back(RegionSuccessor(&getCaseRegion()));
+}
+
+void cir::CaseOp::build(OpBuilder &builder, OperationState &result,
+                        ArrayAttr value, CaseOpKind kind,
+                        OpBuilder::InsertPoint &insertPoint) {
+  OpBuilder::InsertionGuard guardSwitch(builder);
+  result.addAttribute("value", value);
+  result.getOrAddProperties<Properties>().kind =
+      cir::CaseOpKindAttr::get(builder.getContext(), kind);
+  Region *caseRegion = result.addRegion();
+  builder.createBlock(caseRegion);
+
+  insertPoint = builder.saveInsertionPoint();
+}
+
+LogicalResult cir::CaseOp::verify() { return success(); }
----------------
xlauko wrote:

My point was that it is unnecessary, since it has no use returning just 
`success` and can be removed also with  `let hasVerifier = 1;` from operation 
definition. And should be added only when we have some meaningful check in the 
verifier. 

https://github.com/llvm/llvm-project/pull/137106
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits