[mlir][dataflow] Remove early exit in dead code analysis for zero-operand returns

[pengmai]

The PredecessorState in dead code analysis does not register zero-operand returns as predecessors of their corresponding call ops. This causes bugs with dense dataflow analyses that use dead code analysis to query for the predecessors of CallOpInterfaces.

This was reasonable for sparse analyses, but isn't for dense ones.

[llvmbot]

@llvm/pr-subscribers-mlir

Changes

The PredecessorState in dead code analysis does not register zero-operand returns as predecessors of their corresponding call ops. This causes bugs with dense dataflow analyses that use dead code analysis to query for the predecessors of CallOpInterfaces.

---

Full diff: https://github.com/llvm/llvm-project/pull/68151.diff

3 Files Affected:

• (modified) mlir/lib/Analysis/DataFlow/DeadCodeAnalysis.cpp (-4)
• (modified) mlir/test/Analysis/DataFlow/test-dead-code-analysis.mlir (+12)
• (modified) mlir/test/Analysis/DataFlow/test-last-modified-callgraph.mlir (+17)

diff --git a/mlir/lib/Analysis/DataFlow/DeadCodeAnalysis.cpp b/mlir/lib/Analysis/DataFlow/DeadCodeAnalysis.cpp
index d423d37b9770c69..bbe139fd4fb69fe 100644
--- a/mlir/lib/Analysis/DataFlow/DeadCodeAnalysis.cpp
+++ b/mlir/lib/Analysis/DataFlow/DeadCodeAnalysis.cpp
@@ -407,10 +407,6 @@ void DeadCodeAnalysis::visitRegionTerminator(
 
 void DeadCodeAnalysis::visitCallableTerminator(Operation *op,
                                                CallableOpInterface callable) {
-  // If there are no exiting values, we have nothing to do.
-  if (op->getNumOperands() == 0)
-    return;
-
   // Add as predecessors to all callsites this return op.
   auto *callsites = getOrCreateFor<PredecessorState>(op, callable);
   bool canResolve = op->hasTrait<OpTrait::ReturnLike>();
diff --git a/mlir/test/Analysis/DataFlow/test-dead-code-analysis.mlir b/mlir/test/Analysis/DataFlow/test-dead-code-analysis.mlir
index 020c8b2a0e84dc7..b1af670440c2d97 100644
--- a/mlir/test/Analysis/DataFlow/test-dead-code-analysis.mlir
+++ b/mlir/test/Analysis/DataFlow/test-dead-code-analysis.mlir
@@ -118,6 +118,18 @@ func.func @test_callgraph(%cond: i1, %arg0: i32) -> i32 {
   return %2 : i32
 }
 
+func.func private @bax(%arg0: i32) {
+  return {void_return}
+}
+
+func.func @test_callgraph_void_return(%arg0: i32) -> i32 {
+  // CHECK: call_void_return:
+  // CHECK: op_preds: (all) predecessors:
+  // CHECK:   func.return {void_return}
+  func.call @bax(%arg0) {tag = "call_void_return"}: (i32) -> ()
+  return %arg0 : i32
+}
+
 // CHECK: test_unknown_branch:
 // CHECK:  region #0
 // CHECK:   ^bb0 = live
diff --git a/mlir/test/Analysis/DataFlow/test-last-modified-callgraph.mlir b/mlir/test/Analysis/DataFlow/test-last-modified-callgraph.mlir
index 4a243571c231a19..709d787bb306bef 100644
--- a/mlir/test/Analysis/DataFlow/test-last-modified-callgraph.mlir
+++ b/mlir/test/Analysis/DataFlow/test-last-modified-callgraph.mlir
@@ -68,6 +68,23 @@ func.func @test_multiple_return_sites(%cond: i1, %a: i32, %ptr: memref<i32>) ->
 
 // -----
 
+// CHECK-LABEL: test_tag: after_call
+// CHECK: operand #0
+// CHECK-NEXT: - write0
+func.func private @void_return(%ptr: memref<i32>) {
+  return
+}
+
+func.func @test_call_void_return() {
+  %ptr = memref.alloc() : memref<i32>
+  %c0 = arith.constant 0 : i32
+  memref.store %c0, %ptr[] {tag_name = "write0"} : memref<i32>
+  func.call @void_return(%ptr) : (memref<i32>) -> ()
+  memref.load %ptr[] {tag = "after_call"} : memref<i32>
+  return
+}
+
+// -----
 
 func.func private @callee(%arg0: memref<f32>) -> memref<f32> {
   %2 = arith.constant 2.0 : f32