[mlir] Split out a new ControlFlow dialect from Standard

This dialect is intended to model lower level/branch based control-flow constructs. The initial set
of operations are: AssertOp, BranchOp, CondBranchOp, SwitchOp; all split out from the current
standard dialect.

See https://discourse.llvm.org/t/standard-dialect-the-final-chapter/6061

Differential Revision: https://reviews.llvm.org/D118966

Author: River707

flang/include/flang/Optimizer/Support/InitFIR.h

@@ -27,8 +27,8 @@ namespace fir::support {
 #define FLANG_NONCODEGEN_DIALECT_LIST                                          \
   mlir::AffineDialect, FIROpsDialect, mlir::acc::OpenACCDialect,               \
       mlir::omp::OpenMPDialect, mlir::scf::SCFDialect,                         \
-      mlir::arith::ArithmeticDialect, mlir::StandardOpsDialect,                \
-      mlir::vector::VectorDialect
+      mlir::arith::ArithmeticDialect, mlir::cf::ControlFlowDialect,            \
+      mlir::StandardOpsDialect, mlir::vector::VectorDialect
 
 // The definitive list of dialects used by flang.
 #define FLANG_DIALECT_LIST                                                     \

flang/include/flang/Tools/CLOptions.inc

@@ -9,7 +9,7 @@
 /// This file defines some shared command-line options that can be used when
 /// debugging the test tools. This file must be included into the tool.
 
-#include "mlir/Conversion/SCFToStandard/SCFToStandard.h"
+#include "mlir/Conversion/SCFToControlFlow/SCFToControlFlow.h"
 #include "mlir/Pass/PassManager.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 #include "mlir/Transforms/Passes.h"
@@ -139,7 +139,7 @@ inline void createDefaultFIROptimizerPassPipeline(mlir::PassManager &pm) {
 
   // convert control flow to CFG form
   fir::addCfgConversionPass(pm);
-  pm.addPass(mlir::createLowerToCFGPass());
+  pm.addPass(mlir::createConvertSCFToCFPass());
 
   pm.addPass(mlir::createCanonicalizerPass(config));
 }

flang/lib/Lower/CMakeLists.txt

@@ -32,7 +32,7 @@ add_flang_library(FortranLower
   FortranSemantics
   MLIRAffineToStandard
   MLIRLLVMIR
-  MLIRSCFToStandard
+  MLIRSCFToControlFlow
   MLIRStandard
 
   LINK_COMPONENTS

flang/lib/Optimizer/CodeGen/CodeGen.cpp

@@ -18,6 +18,7 @@
 #include "flang/Optimizer/Dialect/FIROps.h"
 #include "flang/Optimizer/Support/TypeCode.h"
 #include "mlir/Conversion/ArithmeticToLLVM/ArithmeticToLLVM.h"
+#include "mlir/Conversion/ControlFlowToLLVM/ControlFlowToLLVM.h"
 #include "mlir/Conversion/LLVMCommon/Pattern.h"
 #include "mlir/Conversion/StandardToLLVM/ConvertStandardToLLVM.h"
 #include "mlir/IR/BuiltinTypes.h"
@@ -3293,6 +3294,8 @@ class FIRToLLVMLowering : public fir::FIRToLLVMLoweringBase<FIRToLLVMLowering> {
     mlir::populateStdToLLVMConversionPatterns(typeConverter, pattern);
     mlir::arith::populateArithmeticToLLVMConversionPatterns(typeConverter,
                                                             pattern);
+    mlir::cf::populateControlFlowToLLVMConversionPatterns(typeConverter,
+                                                          pattern);
     mlir::ConversionTarget target{*context};
     target.addLegalDialect<mlir::LLVM::LLVMDialect>();
 

flang/lib/Optimizer/Transforms/ArrayValueCopy.cpp

@@ -13,6 +13,7 @@
 #include "flang/Optimizer/Dialect/FIRDialect.h"
 #include "flang/Optimizer/Support/FIRContext.h"
 #include "flang/Optimizer/Transforms/Passes.h"
+#include "mlir/Dialect/ControlFlow/IR/ControlFlowOps.h"
 #include "mlir/Dialect/SCF/SCF.h"
 #include "mlir/Transforms/DialectConversion.h"
 #include "llvm/Support/Debug.h"
@@ -332,9 +333,9 @@ ArrayCopyAnalysis::arrayAccesses(ArrayLoadOp load) {
                  << "add modify {" << *owner << "} to array value set\n");
       accesses.push_back(owner);
       appendToQueue(update.getResult(1));
-    } else if (auto br = mlir::dyn_cast<mlir::BranchOp>(owner)) {
+    } else if (auto br = mlir::dyn_cast<mlir::cf::BranchOp>(owner)) {
       branchOp(br.getDest(), br.getDestOperands());
-    } else if (auto br = mlir::dyn_cast<mlir::CondBranchOp>(owner)) {
+    } else if (auto br = mlir::dyn_cast<mlir::cf::CondBranchOp>(owner)) {
       branchOp(br.getTrueDest(), br.getTrueOperands());
       branchOp(br.getFalseDest(), br.getFalseOperands());
     } else if (mlir::isa<ArrayMergeStoreOp>(owner)) {
@@ -789,9 +790,9 @@ class ArrayValueCopyConverter
     patterns1.insert<ArrayUpdateConversion>(context, analysis, useMap);
     patterns1.insert<ArrayModifyConversion>(context, analysis, useMap);
     mlir::ConversionTarget target(*context);
-    target.addLegalDialect<FIROpsDialect, mlir::scf::SCFDialect,
-                           mlir::arith::ArithmeticDialect,
-                           mlir::StandardOpsDialect>();
+    target.addLegalDialect<
+        FIROpsDialect, mlir::scf::SCFDialect, mlir::arith::ArithmeticDialect,
+        mlir::cf::ControlFlowDialect, mlir::StandardOpsDialect>();
     target.addIllegalOp<ArrayFetchOp, ArrayUpdateOp, ArrayModifyOp>();
     // Rewrite the array fetch and array update ops.
     if (mlir::failed(

flang/lib/Optimizer/Transforms/RewriteLoop.cpp

@@ -11,6 +11,7 @@
 #include "flang/Optimizer/Dialect/FIROps.h"
 #include "flang/Optimizer/Transforms/Passes.h"
 #include "mlir/Dialect/Affine/IR/AffineOps.h"
+#include "mlir/Dialect/ControlFlow/IR/ControlFlowOps.h"
 #include "mlir/Dialect/StandardOps/IR/Ops.h"
 #include "mlir/Pass/Pass.h"
 #include "mlir/Transforms/DialectConversion.h"
@@ -84,7 +85,7 @@ class CfgLoopConv : public mlir::OpRewritePattern<fir::DoLoopOp> {
     loopOperands.append(operands.begin(), operands.end());
     loopOperands.push_back(iters);
 
-    rewriter.create<mlir::BranchOp>(loc, conditionalBlock, loopOperands);
+    rewriter.create<mlir::cf::BranchOp>(loc, conditionalBlock, loopOperands);
 
     // Last loop block
     auto *terminator = lastBlock->getTerminator();
@@ -105,7 +106,7 @@ class CfgLoopConv : public mlir::OpRewritePattern<fir::DoLoopOp> {
                                    : terminator->operand_begin();
     loopCarried.append(begin, terminator->operand_end());
     loopCarried.push_back(itersMinusOne);
-    rewriter.create<mlir::BranchOp>(loc, conditionalBlock, loopCarried);
+    rewriter.create<mlir::cf::BranchOp>(loc, conditionalBlock, loopCarried);
     rewriter.eraseOp(terminator);
 
     // Conditional block
@@ -114,9 +115,9 @@ class CfgLoopConv : public mlir::OpRewritePattern<fir::DoLoopOp> {
     auto comparison = rewriter.create<mlir::arith::CmpIOp>(
         loc, arith::CmpIPredicate::sgt, itersLeft, zero);
 
-    rewriter.create<mlir::CondBranchOp>(loc, comparison, firstBlock,
-                                        llvm::ArrayRef<mlir::Value>(), endBlock,
-                                        llvm::ArrayRef<mlir::Value>());
+    rewriter.create<mlir::cf::CondBranchOp>(
+        loc, comparison, firstBlock, llvm::ArrayRef<mlir::Value>(), endBlock,
+        llvm::ArrayRef<mlir::Value>());
 
     // The result of the loop operation is the values of the condition block
     // arguments except the induction variable on the last iteration.
@@ -155,7 +156,7 @@ class CfgIfConv : public mlir::OpRewritePattern<fir::IfOp> {
     } else {
       continueBlock =
           rewriter.createBlock(remainingOpsBlock, ifOp.getResultTypes());
-      rewriter.create<mlir::BranchOp>(loc, remainingOpsBlock);
+      rewriter.create<mlir::cf::BranchOp>(loc, remainingOpsBlock);
     }
 
     // Move blocks from the "then" region to the region containing 'fir.if',
@@ -165,7 +166,8 @@ class CfgIfConv : public mlir::OpRewritePattern<fir::IfOp> {
     auto *ifOpTerminator = ifOpRegion.back().getTerminator();
     auto ifOpTerminatorOperands = ifOpTerminator->getOperands();
     rewriter.setInsertionPointToEnd(&ifOpRegion.back());
-    rewriter.create<mlir::BranchOp>(loc, continueBlock, ifOpTerminatorOperands);
+    rewriter.create<mlir::cf::BranchOp>(loc, continueBlock,
+                                        ifOpTerminatorOperands);
     rewriter.eraseOp(ifOpTerminator);
     rewriter.inlineRegionBefore(ifOpRegion, continueBlock);
 
@@ -179,14 +181,14 @@ class CfgIfConv : public mlir::OpRewritePattern<fir::IfOp> {
       auto *otherwiseTerm = otherwiseRegion.back().getTerminator();
       auto otherwiseTermOperands = otherwiseTerm->getOperands();
       rewriter.setInsertionPointToEnd(&otherwiseRegion.back());
-      rewriter.create<mlir::BranchOp>(loc, continueBlock,
-                                      otherwiseTermOperands);
+      rewriter.create<mlir::cf::BranchOp>(loc, continueBlock,
+                                          otherwiseTermOperands);
       rewriter.eraseOp(otherwiseTerm);
       rewriter.inlineRegionBefore(otherwiseRegion, continueBlock);
     }
 
     rewriter.setInsertionPointToEnd(condBlock);
-    rewriter.create<mlir::CondBranchOp>(
+    rewriter.create<mlir::cf::CondBranchOp>(
         loc, ifOp.condition(), ifOpBlock, llvm::ArrayRef<mlir::Value>(),
         otherwiseBlock, llvm::ArrayRef<mlir::Value>());
     rewriter.replaceOp(ifOp, continueBlock->getArguments());
@@ -241,7 +243,7 @@ class CfgIterWhileConv : public mlir::OpRewritePattern<fir::IterWhileOp> {
     auto begin = whileOp.finalValue() ? std::next(terminator->operand_begin())
                                       : terminator->operand_begin();
     loopCarried.append(begin, terminator->operand_end());
-    rewriter.create<mlir::BranchOp>(loc, conditionBlock, loopCarried);
+    rewriter.create<mlir::cf::BranchOp>(loc, conditionBlock, loopCarried);
     rewriter.eraseOp(terminator);
 
     // Compute loop bounds before branching to the condition.
@@ -256,7 +258,7 @@ class CfgIterWhileConv : public mlir::OpRewritePattern<fir::IterWhileOp> {
     destOperands.push_back(lowerBound);
     auto iterOperands = whileOp.getIterOperands();
     destOperands.append(iterOperands.begin(), iterOperands.end());
-    rewriter.create<mlir::BranchOp>(loc, conditionBlock, destOperands);
+    rewriter.create<mlir::cf::BranchOp>(loc, conditionBlock, destOperands);
 
     // With the body block done, we can fill in the condition block.
     rewriter.setInsertionPointToEnd(conditionBlock);
@@ -278,9 +280,9 @@ class CfgIterWhileConv : public mlir::OpRewritePattern<fir::IterWhileOp> {
     // Remember to AND in the early-exit bool.
     auto comparison =
         rewriter.create<mlir::arith::AndIOp>(loc, iterateVar, cmp2);
-    rewriter.create<mlir::CondBranchOp>(loc, comparison, firstBodyBlock,
-                                        llvm::ArrayRef<mlir::Value>(), endBlock,
-                                        llvm::ArrayRef<mlir::Value>());
+    rewriter.create<mlir::cf::CondBranchOp>(
+        loc, comparison, firstBodyBlock, llvm::ArrayRef<mlir::Value>(),
+        endBlock, llvm::ArrayRef<mlir::Value>());
     // The result of the loop operation is the values of the condition block
     // arguments except the induction variable on the last iteration.
     auto args = whileOp.finalValue()
@@ -300,8 +302,8 @@ class CfgConversion : public CFGConversionBase<CfgConversion> {
     patterns.insert<CfgLoopConv, CfgIfConv, CfgIterWhileConv>(
         context, forceLoopToExecuteOnce);
     mlir::ConversionTarget target(*context);
-    target.addLegalDialect<mlir::AffineDialect, FIROpsDialect,
-                           mlir::StandardOpsDialect>();
+    target.addLegalDialect<mlir::AffineDialect, mlir::cf::ControlFlowDialect,
+                           FIROpsDialect, mlir::StandardOpsDialect>();
 
     // apply the patterns
     target.addIllegalOp<ResultOp, DoLoopOp, IfOp, IterWhileOp>();

flang/test/Fir/Todo/select_case_with_character.fir

@@ -10,10 +10,10 @@ func @select_case_charachter(%arg0: !fir.char<2, 10>, %arg1: !fir.char<2, 10>, %
                                             unit, ^bb3]
 ^bb1:
   %c1_i32 = arith.constant 1 : i32
-  br ^bb3
+  cf.br ^bb3
 ^bb2:
   %c2_i32 = arith.constant 2 : i32
-  br ^bb3
+  cf.br ^bb3
 ^bb3:
   return
 }

flang/test/Fir/convert-to-llvm.fir

@@ -1175,23 +1175,23 @@ func @select_case_integer(%arg0: !fir.ref<i32>) -> i32 {
 ^bb1:  // pred: ^bb0
   %c1_i32_0 = arith.constant 1 : i32
   fir.store %c1_i32_0 to %arg0 : !fir.ref<i32>
-  br ^bb6
+  cf.br ^bb6
 ^bb2:  // pred: ^bb0
   %c2_i32_1 = arith.constant 2 : i32
   fir.store %c2_i32_1 to %arg0 : !fir.ref<i32>
-  br ^bb6
+  cf.br ^bb6
 ^bb3:  // pred: ^bb0
   %c0_i32 = arith.constant 0 : i32
   fir.store %c0_i32 to %arg0 : !fir.ref<i32>
-  br ^bb6
+  cf.br ^bb6
 ^bb4:  // pred: ^bb0
   %c4_i32_2 = arith.constant 4 : i32
   fir.store %c4_i32_2 to %arg0 : !fir.ref<i32>
-  br ^bb6
+  cf.br ^bb6
 ^bb5:  // 3 preds: ^bb0, ^bb0, ^bb0
   %c7_i32_3 = arith.constant 7 : i32
   fir.store %c7_i32_3 to %arg0 : !fir.ref<i32>
-  br ^bb6
+  cf.br ^bb6
 ^bb6:  // 5 preds: ^bb1, ^bb2, ^bb3, ^bb4, ^bb5
   %3 = fir.load %arg0 : !fir.ref<i32>
   return %3 : i32
@@ -1275,10 +1275,10 @@ func @select_case_logical(%arg0: !fir.ref<!fir.logical<4>>) {
                             unit, ^bb3]
 ^bb1:
   %c1_i32 = arith.constant 1 : i32
-  br ^bb3
+  cf.br ^bb3
 ^bb2:
   %c2_i32 = arith.constant 2 : i32
-  br ^bb3
+  cf.br ^bb3
 ^bb3:
   return
 }

flang/test/Fir/memref-data-flow.fir

@@ -9,10 +9,10 @@ func @load_store_chain_removal(%arg0: !fir.ref<!fir.array<60xi32>>, %arg1: !fir.
   %c1 = arith.constant 1 : index
   %0 = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFf1dcEi"}
   %1 = fir.alloca !fir.array<60xi32> {bindc_name = "t1", uniq_name = "_QFf1dcEt1"}
-  br ^bb1(%c1, %c60 : index, index)
+  cf.br ^bb1(%c1, %c60 : index, index)
 ^bb1(%2: index, %3: index):  // 2 preds: ^bb0, ^bb2
   %4 = arith.cmpi sgt, %3, %c0 : index
-  cond_br %4, ^bb2, ^bb3
+  cf.cond_br %4, ^bb2, ^bb3
 ^bb2:  // pred: ^bb1
   %5 = fir.convert %2 : (index) -> i32
   fir.store %5 to %0 : !fir.ref<i32>
@@ -26,14 +26,14 @@ func @load_store_chain_removal(%arg0: !fir.ref<!fir.array<60xi32>>, %arg1: !fir.
   fir.store %11 to %12 : !fir.ref<i32>
   %13 = arith.addi %2, %c1 : index
   %14 = arith.subi %3, %c1 : index
-  br ^bb1(%13, %14 : index, index)
+  cf.br ^bb1(%13, %14 : index, index)
 ^bb3:  // pred: ^bb1
   %15 = fir.convert %2 : (index) -> i32
   fir.store %15 to %0 : !fir.ref<i32>
-  br ^bb4(%c1, %c60 : index, index)
+  cf.br ^bb4(%c1, %c60 : index, index)
 ^bb4(%16: index, %17: index):  // 2 preds: ^bb3, ^bb5
   %18 = arith.cmpi sgt, %17, %c0 : index
-  cond_br %18, ^bb5, ^bb6
+  cf.cond_br %18, ^bb5, ^bb6
 ^bb5:  // pred: ^bb4
   %19 = fir.convert %16 : (index) -> i32
   fir.store %19 to %0 : !fir.ref<i32>
@@ -49,7 +49,7 @@ func @load_store_chain_removal(%arg0: !fir.ref<!fir.array<60xi32>>, %arg1: !fir.
   fir.store %27 to %28 : !fir.ref<i32>
   %29 = arith.addi %16, %c1 : index
   %30 = arith.subi %17, %c1 : index
-  br ^bb4(%29, %30 : index, index)
+  cf.br ^bb4(%29, %30 : index, index)
 ^bb6:  // pred: ^bb4
   %31 = fir.convert %16 : (index) -> i32
   fir.store %31 to %0 : !fir.ref<i32>

flang/tools/bbc/CMakeLists.txt

@@ -13,7 +13,7 @@ FIRTransforms
 FIRBuilder
 ${dialect_libs}
 MLIRAffineToStandard
-MLIRSCFToStandard
+MLIRSCFToControlFlow
 FortranCommon
 FortranParser
 FortranEvaluate

flang/tools/bbc/bbc.cpp

@@ -38,7 +38,6 @@
 #include "flang/Semantics/semantics.h"
 #include "flang/Semantics/unparse-with-symbols.h"
 #include "flang/Version.inc"
-#include "mlir/Conversion/SCFToStandard/SCFToStandard.h"
 #include "mlir/IR/AsmState.h"
 #include "mlir/IR/BuiltinOps.h"
 #include "mlir/IR/MLIRContext.h"

flang/tools/fir-opt/CMakeLists.txt

@@ -18,7 +18,7 @@ target_link_libraries(fir-opt PRIVATE
   MLIRTransforms
   MLIRAffineToStandard
   MLIRAnalysis
-  MLIRSCFToStandard
+  MLIRSCFToControlFlow
   MLIRParser
   MLIRStandardToLLVM
   MLIRSupport

flang/tools/tco/CMakeLists.txt

@@ -17,7 +17,7 @@ target_link_libraries(tco PRIVATE
   MLIRTransforms
   MLIRAffineToStandard
   MLIRAnalysis
-  MLIRSCFToStandard
+  MLIRSCFToControlFlow
   MLIRParser
   MLIRStandardToLLVM
   MLIRSupport

flang/tools/tco/tco.cpp

@@ -17,7 +17,6 @@
 #include "flang/Optimizer/Support/InternalNames.h"
 #include "flang/Optimizer/Support/KindMapping.h"
 #include "flang/Optimizer/Transforms/Passes.h"
-#include "mlir/Conversion/SCFToStandard/SCFToStandard.h"
 #include "mlir/IR/AsmState.h"
 #include "mlir/IR/BuiltinOps.h"
 #include "mlir/IR/MLIRContext.h"

mlir/benchmark/python/common.py

@@ -26,7 +26,7 @@ def setup_passes(mlir_module):
         f"sparse-tensor-conversion,"
         f"builtin.func"
         f"(linalg-bufferize,convert-linalg-to-loops,convert-vector-to-scf),"
-        f"convert-scf-to-std,"
+        f"convert-scf-to-cf,"
         f"func-bufferize,"
         f"arith-bufferize,"
         f"builtin.func(tensor-bufferize,finalizing-bufferize),"