[flang-commits] [flang] [NFC][mlir][OpenMP] Remove mentions of `target` from generic `loop` rewrite (PR #124528)

[Kareem Ergawy via flang-commits]

https://github.com/ergawy updated https://github.com/llvm/llvm-project/pull/124528

>From 7c4ec9b89b8af36cf49101c654ae8cd7e8fd4418 Mon Sep 17 00:00:00 2001
From: ergawy <kareem.ergawy at amd.com>
Date: Mon, 27 Jan 2025 04:56:13 -0600
Subject: [PATCH] [NFC][mlir][OpenMP] Remove mentions of `target` from generic
 loop rewrite

This removes mentions of `target` from the generic `loop` rewrite pass
since there is not need for it anyway. It is enough to detect `loop`'s
nesting within `teams` or `parallel` directives.
---
 .../OpenMP/GenericLoopConversion.cpp          | 24 +++++++------------
 .../generic-loop-rewriting-todo.mlir          |  2 +-
 2 files changed, 10 insertions(+), 16 deletions(-)

diff --git a/flang/lib/Optimizer/OpenMP/GenericLoopConversion.cpp b/flang/lib/Optimizer/OpenMP/GenericLoopConversion.cpp
index c95d625d7240b4..546943204e6534 100644
--- a/flang/lib/Optimizer/OpenMP/GenericLoopConversion.cpp
+++ b/flang/lib/Optimizer/OpenMP/GenericLoopConversion.cpp
@@ -29,11 +29,7 @@ namespace {
 class GenericLoopConversionPattern
     : public mlir::OpConversionPattern<mlir::omp::LoopOp> {
 public:
-  enum class GenericLoopCombinedInfo {
-    Standalone,
-    TargetTeamsLoop,
-    TargetParallelLoop
-  };
+  enum class GenericLoopCombinedInfo { Standalone, TeamsLoop, ParallelLoop };
 
   using mlir::OpConversionPattern<mlir::omp::LoopOp>::OpConversionPattern;
 
@@ -55,10 +51,10 @@ class GenericLoopConversionPattern
     case GenericLoopCombinedInfo::Standalone:
       rewriteStandaloneLoop(loopOp, rewriter);
       break;
-    case GenericLoopCombinedInfo::TargetParallelLoop:
+    case GenericLoopCombinedInfo::ParallelLoop:
       llvm_unreachable("not yet implemented: `parallel loop` direcitve");
       break;
-    case GenericLoopCombinedInfo::TargetTeamsLoop:
+    case GenericLoopCombinedInfo::TeamsLoop:
       rewriteToDistributeParallelDo(loopOp, rewriter);
       break;
     }
@@ -74,10 +70,10 @@ class GenericLoopConversionPattern
     switch (combinedInfo) {
     case GenericLoopCombinedInfo::Standalone:
       break;
-    case GenericLoopCombinedInfo::TargetParallelLoop:
+    case GenericLoopCombinedInfo::ParallelLoop:
       return loopOp.emitError(
-          "not yet implemented: Combined `omp target parallel loop` directive");
-    case GenericLoopCombinedInfo::TargetTeamsLoop:
+          "not yet implemented: Combined `parallel loop` directive");
+    case GenericLoopCombinedInfo::TeamsLoop:
       break;
     }
 
@@ -99,7 +95,7 @@ class GenericLoopConversionPattern
     if (!loopOp.getReductionVars().empty())
       return todo("reduction");
 
-    // TODO For `target teams loop`, check similar constrains to what is checked
+    // TODO For `teams loop`, check similar constrains to what is checked
     // by `TeamsLoopChecker` in SemaOpenMP.cpp.
     return mlir::success();
   }
@@ -111,13 +107,11 @@ class GenericLoopConversionPattern
     GenericLoopCombinedInfo result = GenericLoopCombinedInfo::Standalone;
 
     if (auto teamsOp = mlir::dyn_cast_if_present<mlir::omp::TeamsOp>(parentOp))
-      if (mlir::isa_and_present<mlir::omp::TargetOp>(teamsOp->getParentOp()))
-        result = GenericLoopCombinedInfo::TargetTeamsLoop;
+      result = GenericLoopCombinedInfo::TeamsLoop;
 
     if (auto parallelOp =
             mlir::dyn_cast_if_present<mlir::omp::ParallelOp>(parentOp))
-      if (mlir::isa_and_present<mlir::omp::TargetOp>(parallelOp->getParentOp()))
-        result = GenericLoopCombinedInfo::TargetParallelLoop;
+      result = GenericLoopCombinedInfo::ParallelLoop;
 
     return result;
   }
diff --git a/flang/test/Transforms/generic-loop-rewriting-todo.mlir b/flang/test/Transforms/generic-loop-rewriting-todo.mlir
index becd6b8dcb5cb4..3259ceca70d50d 100644
--- a/flang/test/Transforms/generic-loop-rewriting-todo.mlir
+++ b/flang/test/Transforms/generic-loop-rewriting-todo.mlir
@@ -6,7 +6,7 @@ func.func @_QPtarget_parallel_loop() {
       %c0 = arith.constant 0 : i32
       %c10 = arith.constant 10 : i32
       %c1 = arith.constant 1 : i32
-      // expected-error at below {{not yet implemented: Combined `omp target parallel loop` directive}}
+      // expected-error at below {{not yet implemented: Combined `parallel loop` directive}}
       omp.loop {
         omp.loop_nest (%arg3) : i32 = (%c0) to (%c10) inclusive step (%c1) {
           omp.yield