[flang][fir] Add fir.if -> scf.if and add filecheck test file

[StarryCSF]

This commmit is a supplement for #140374.
RFC:https://discourse.llvm.org/t/rfc-add-fir-affine-optimization-fir-pass-pipeline/86190/6

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-flang-fir-hlfir

Author: Q (StarryCSF)

Changes

This commmit is a supplement for #140374.
RFC:https://discourse.llvm.org/t/rfc-add-fir-affine-optimization-fir-pass-pipeline/86190/6

---

Full diff: https://github.com/llvm/llvm-project/pull/142965.diff

2 Files Affected:

• (modified) flang/lib/Optimizer/Transforms/FIRToSCF.cpp (+56-2)
• (added) flang/test/Fir/FirToSCF/if.fir (+97)

diff --git a/flang/lib/Optimizer/Transforms/FIRToSCF.cpp b/flang/lib/Optimizer/Transforms/FIRToSCF.cpp
index f06ad2db90d55..12a6d3aad4c0d 100644
--- a/flang/lib/Optimizer/Transforms/FIRToSCF.cpp
+++ b/flang/lib/Optimizer/Transforms/FIRToSCF.cpp
@@ -87,13 +87,67 @@ struct DoLoopConversion : public OpRewritePattern<fir::DoLoopOp> {
     return success();
   }
 };
+
+struct IfConversion : public OpRewritePattern<fir::IfOp> {
+  using OpRewritePattern<fir::IfOp>::OpRewritePattern;
+  LogicalResult matchAndRewrite(fir::IfOp ifOp,
+                                PatternRewriter &rewriter) const override {
+    mlir::Location loc = ifOp.getLoc();
+    mlir::detail::TypedValue<mlir::IntegerType> condition = ifOp.getCondition();
+    ValueTypeRange<ResultRange> resultTypes = ifOp.getResultTypes();
+    bool hasResult = !resultTypes.empty();
+    auto scfIfOp = rewriter.create<scf::IfOp>(loc, resultTypes, condition,
+                                              !ifOp.getElseRegion().empty());
+    // then region
+    assert(!ifOp.getThenRegion().empty() && "must have then region");
+    auto &firThenBlock = ifOp.getThenRegion().front();
+    auto &scfThenBlock = scfIfOp.getThenRegion().front();
+    auto &firThenOps = firThenBlock.getOperations();
+    mlir::Operation *firThenTerminator = firThenBlock.getTerminator();
+
+    rewriter.setInsertionPointToStart(&scfThenBlock);
+    // not splice terminator
+    scfThenBlock.getOperations().splice(scfThenBlock.begin(), firThenOps,
+                                        firThenOps.begin(),
+                                        std::prev(firThenOps.end()));
+    // create terminator scf.yield
+    if (hasResult) {
+      rewriter.setInsertionPointToEnd(&scfThenBlock);
+      mlir::OperandRange thenResults = firThenTerminator->getOperands();
+      rewriter.create<scf::YieldOp>(firThenTerminator->getLoc(), thenResults);
+    }
+
+    // else region
+    if (!ifOp.getElseRegion().empty()) {
+      auto &firElseBlock = ifOp.getElseRegion().front();
+      auto &scfElseBlock = scfIfOp.getElseRegion().front();
+      auto &firElseOps = firElseBlock.getOperations();
+      mlir::Operation *firElseTerminator = firElseBlock.getTerminator();
+
+      rewriter.setInsertionPointToStart(&scfElseBlock);
+      scfElseBlock.getOperations().splice(scfElseBlock.begin(), firElseOps,
+                                          firElseOps.begin(),
+                                          std::prev(firElseOps.end()));
+
+      if (hasResult) {
+        rewriter.setInsertionPointToEnd(&scfElseBlock);
+        mlir::OperandRange elseResults = firElseTerminator->getOperands();
+        rewriter.create<scf::YieldOp>(firElseTerminator->getLoc(), elseResults);
+      }
+    }
+
+    scfIfOp->setAttrs(ifOp->getAttrs());
+    rewriter.replaceOp(ifOp, scfIfOp);
+    return success();
+  }
+};
 } // namespace
 
 void FIRToSCFPass::runOnOperation() {
   RewritePatternSet patterns(&getContext());
-  patterns.add<DoLoopConversion>(patterns.getContext());
+  patterns.add<DoLoopConversion, IfConversion>(patterns.getContext());
   ConversionTarget target(getContext());
-  target.addIllegalOp<fir::DoLoopOp>();
+  target.addIllegalOp<fir::DoLoopOp, fir::IfOp>();
   target.markUnknownOpDynamicallyLegal([](Operation *) { return true; });
   if (failed(
           applyPartialConversion(getOperation(), target, std::move(patterns))))
diff --git a/flang/test/Fir/FirToSCF/if.fir b/flang/test/Fir/FirToSCF/if.fir
new file mode 100644
index 0000000000000..8885b0343095d
--- /dev/null
+++ b/flang/test/Fir/FirToSCF/if.fir
@@ -0,0 +1,97 @@
+
+// RUN: fir-opt %s --fir-to-scf | FileCheck %s
+
+// CHECK: func.func @_QFPtest_only(
+// CHECK:   %[[ARG0:.*]]: !fir.ref<tuple<!fir.ref<f32>>>) {
+// CHECK:   %[[VAL_0:.*]] = arith.constant 1 : i32
+// CHECK:   %[[VAL_1:.*]] = arith.constant false
+// CHECK:   %[[VAL_2:.*]] = arith.constant 0 : i32
+// CHECK:   scf.if %[[VAL_1:.*]] {
+// CHECK:    %[[VAL_3:.*]] = arith.addi %[[VAL_2:.*]], %[[VAL_0:.*]] : i32
+// CHECK:   }
+// CHECK:   return
+// CHECK:   }
+func.func @_QFPtest_only(%arg0: !fir.ref<tuple<!fir.ref<f32>>>) {
+    %c1_i32 = arith.constant 1 : i32
+    %false = arith.constant false
+    %c0_i32 = arith.constant 0 : i32
+    fir.if %false {
+      %0 = arith.addi %c0_i32, %c1_i32 : i32
+    }
+    return
+  }
+
+// CHECK: func.func @_QFPtest_else(
+// CHECK:   %[[ARG0:.*]]: !fir.ref<tuple<!fir.ref<f32>>>) {
+// CHECK:   %[[VAL_0:.*]] = arith.constant 1 : i32
+// CHECK:   %[[VAL_1:.*]] = arith.constant false
+// CHECK:   %[[VAL_2:.*]] = arith.constant 0 : i32
+// CHECK:   %[[VAL_3:.*]] = fir.dummy_scope : !fir.dscope
+// CHECK:   %[[VAL_4:.*]] = fir.coordinate_of %[[ARG0:.*]], %[[VAL_2:.*]] : (!fir.ref<tuple<!fir.ref<f32>>>, i32) -> !fir.llvm_ptr<!fir.ref<f32>>
+// CHECK:   %[[VAL_5:.*]] = fir.load %[[VAL_4:.*]] : !fir.llvm_ptr<!fir.ref<f32>>
+// CHECK:   %[[VAL_6:.*]] = fir.declare %[[VAL_5:.*]] {fortran_attrs = #fir.var_attrs<host_assoc>, uniq_name = "_QFEx"} : (!fir.ref<f32>) -> !fir.ref<f32>
+// CHECK:   %[[VAL_7:.*]] = fir.address_of(@_QFFtest_elseEsum) : !fir.ref<i32>
+// CHECK:   %[[VAL_10:.*]] = fir.declare %[[VAL_7:.*]] {uniq_name = "_QFFtest_elseEsum"} : (!fir.ref<i32>) -> !fir.ref<i32>
+// CHECK:   scf.if %[[VAL_1:.*]] {
+// CHECK:     %[[VAL_8:.*]] = fir.load %[[VAL_10:.*]] : !fir.ref<i32>
+// CHECK:     %[[VAL_9:.*]] = arith.addi %[[VAL_8:.*]], %[[VAL_0:.*]] : i32
+// CHECK:     fir.store %[[VAL_9:.*]] to %[[VAL_10:.*]] : !fir.ref<i32>
+// CHECK:   } else {
+// CHECK:     %[[VAL_8:.*]] = fir.load %[[VAL_10:.*]] : !fir.ref<i32>
+// CHECK:     %[[VAL_9:.*]] = arith.addi %[[VAL_8:.*]], %c1_i32 : i32
+// CHECK:     fir.store %[[VAL_9:.*]] to %[[VAL_10:.*]] : !fir.ref<i32>
+// CHECK:   }
+// CHECK:   return
+// CHECK: }
+func.func  @_QFPtest_else(%arg0: !fir.ref<tuple<!fir.ref<f32>>> {}) attributes {} {
+  %c1_i32 = arith.constant 1 : i32
+  %false = arith.constant false
+  %c0_i32 = arith.constant 0 : i32
+  %0 = fir.dummy_scope : !fir.dscope
+  %1 = fir.coordinate_of %arg0, %c0_i32 : (!fir.ref<tuple<!fir.ref<f32>>>, i32) -> !fir.llvm_ptr<!fir.ref<f32>>
+  %2 = fir.load %1 : !fir.llvm_ptr<!fir.ref<f32>>
+  %3 = fir.declare %2 {fortran_attrs = #fir.var_attrs<host_assoc>, uniq_name = "_QFEx"} : (!fir.ref<f32>) -> !fir.ref<f32>
+  %4 = fir.address_of(@_QFFtest_elseEsum) : !fir.ref<i32>
+  %5 = fir.declare %4 {uniq_name = "_QFFtest_elseEsum"} : (!fir.ref<i32>) -> !fir.ref<i32>
+  fir.if %false {
+    %6 = fir.load %5 : !fir.ref<i32>
+    %7 = arith.addi %6, %c1_i32 : i32
+    fir.store %7 to %5 : !fir.ref<i32>
+  } else {
+    %6 = fir.load %5 : !fir.ref<i32>
+    %7 = arith.addi %6, %c1_i32 : i32
+    fir.store %7 to %5 : !fir.ref<i32>
+  }
+  return
+}
+
+// CHECK-LABEL:     func.func @test_two_result() {
+// CHECK:           %[[VAL_0:.*]] = arith.constant 10 : i32
+// CHECK:           %[[VAL_1:.*]] = arith.constant 5 : i32
+// CHECK:           %[[VAL_2:.*]] = arith.cmpi sgt, %[[VAL_0:.*]], %[[VAL_1:.*]] : i32
+// CHECK:           %[[VAL_3:.*]] = arith.constant 3.140000e+00 : f32
+// CHECK:           %[[VAL_4:.*]] = arith.constant 2.710000e+00 : f32
+// CHECK:           %[[VAL_5:.*]] = arith.constant 1.000000e+00 : f32
+// CHECK:           %[[VAL_6:.*]] = arith.constant 2.000000e+00 : f32
+// CHECK:           %[[RES:[0-9]+]]:2 = scf.if %[[VAL_2:.*]] -> (f32, f32) {
+// CHECK:            scf.yield %[[VAL_3:.*]], %[[VAL_4:.*]] : f32, f32
+// CHECK:            } else {
+// CHECK:              scf.yield %[[VAL_5:.*]], %[[VAL_6:.*]] : f32, f32
+// CHECK:            }
+// CHECK:            return 
+// CHECK:            }
+func.func @test_two_result() {
+  %c10_i32 = arith.constant 10 : i32
+  %c5_i32 = arith.constant 5 : i32
+  %cmp = arith.cmpi sgt, %c10_i32, %c5_i32 : i32
+  %c3_14_f32 = arith.constant 3.14 : f32
+  %c2_71_f32 = arith.constant 2.71 : f32
+  %c1_0_f32 = arith.constant 1.0 : f32
+  %c2_0_f32 = arith.constant 2.0 : f32
+  %x, %y = fir.if %cmp -> (f32, f32) {
+    fir.result %c3_14_f32, %c2_71_f32 : f32, f32
+  } else {
+    fir.result %c1_0_f32, %c2_0_f32 : f32, f32
+  }
+  return
+}

[tblah]

See my comment on the old PR

[NexMing]

LGTM

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[rscottmanley]

LGTM

[vzakhari]

Thank you for the update!

[github-actions]

@StarryCSF Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

[lei137]

Sorry. I need to revert this patch since the builder flang-ppc64 have been red for 4 days due to the new tc segfaults on it:

$ fir-opt  --fir-to-scf flang/test/Fir/FirToSCF/if.fir 
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.
Stack dump:
0.      Program arguments: fir-opt --fir-to-scf flang/test/Fir/FirToSCF/if.fir --fir-to-scf
Segmentation fault (core dumped)