[flang][HLFIR] Use GreedyPatternRewriter in LowerHLFIRIntrinsics #83438
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
In llvm#83253 @matthias-springer pointed out that LowerHLFIRIntrinsics.cpp should not be using rewrite patterns with the dialect conversion driver. The intention of this pass is to lower HLFIR intrinsic operations into FIR so it conceptually fits dialect conversion. However, dialect conversion is much stricter about changing types when replacing operations. This pass sometimes looses track of array bounds, resulting in replacements with operations with different but compatible types (expressions of the same rank and element types but with or without compile time known array bounds). This is difficult to accommodate with the dialect conversion driver and so I have changed to use the greedy pattern rewriter. There is a lot of test churn because the greedy pattern rewriter also performs canonicalization.
tblah
requested review from
matthias-springer,
kiranchandramohan and
davemgreen
tblah
changed the title
llvmbot
added
flang
|
@llvm/pr-subscribers-flang-fir-hlfir Author: Tom Eccles (tblah) ChangesIn #83253 @matthias-springer pointed out that LowerHLFIRIntrinsics.cpp should not be using rewrite patterns with the dialect conversion driver. The intention of this pass is to lower HLFIR intrinsic operations into FIR so it conceptually fits dialect conversion. However, dialect conversion is much stricter about changing types when replacing operations. This pass sometimes looses track of array bounds, resulting in replacements with operations with different but compatible types (expressions of the same rank and element types but with or without compile time known array bounds). This is difficult to accommodate with the dialect conversion driver and so I have changed to use the greedy pattern rewriter. There is a lot of test churn because the greedy pattern rewriter also performs canonicalization. Patch is 99.64 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/83438.diff 17 Files Affected:
diff --git a/flang/lib/Optimizer/HLFIR/Transforms/LowerHLFIRIntrinsics.cpp b/flang/lib/Optimizer/HLFIR/Transforms/LowerHLFIRIntrinsics.cpp
index 377cc44392028f..0142fb0cfb0bb0 100644
--- a/flang/lib/Optimizer/HLFIR/Transforms/LowerHLFIRIntrinsics.cpp
+++ b/flang/lib/Optimizer/HLFIR/Transforms/LowerHLFIRIntrinsics.cpp
@@ -18,12 +18,12 @@
#include "flang/Optimizer/HLFIR/HLFIROps.h"
#include "flang/Optimizer/HLFIR/Passes.h"
#include "mlir/IR/BuiltinDialect.h"
+#include "mlir/IR/MLIRContext.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/Pass/Pass.h"
#include "mlir/Pass/PassManager.h"
#include "mlir/Support/LogicalResult.h"
-#include "mlir/Transforms/DialectConversion.h"
-#include <mlir/IR/MLIRContext.h>
+#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
#include <optional>
namespace hlfir {
@@ -176,14 +176,7 @@ class HlfirIntrinsicConversion : public mlir::OpRewritePattern<OP> {
rewriter.eraseOp(use);
}
}
- // TODO: This entire pass should be a greedy pattern rewrite or a manual
- // IR traversal. A dialect conversion cannot be used here because
- // `replaceAllUsesWith` is not supported. Similarly, `replaceOp` is not
- // suitable because "op->getResult(0)" and "base" can have different types.
- // In such a case, the dialect conversion will attempt to convert the type,
- // but no type converter is specified in this pass. Also note that all
- // patterns in this pass are actually rewrite patterns.
- op->getResult(0).replaceAllUsesWith(base);
+
rewriter.replaceOp(op, base);
}
};
@@ -491,19 +484,19 @@ class LowerHLFIRIntrinsics
ProductOpConversion, TransposeOpConversion, CountOpConversion,
DotProductOpConversion, MaxvalOpConversion, MinvalOpConversion,
MinlocOpConversion, MaxlocOpConversion>(context);
- mlir::ConversionTarget target(*context);
- target.addLegalDialect<mlir::BuiltinDialect, mlir::arith::ArithDialect,
- mlir::func::FuncDialect, fir::FIROpsDialect,
- hlfir::hlfirDialect>();
- target.addIllegalOp<hlfir::MatmulOp, hlfir::MatmulTransposeOp, hlfir::SumOp,
- hlfir::ProductOp, hlfir::TransposeOp, hlfir::AnyOp,
- hlfir::AllOp, hlfir::DotProductOp, hlfir::CountOp,
- hlfir::MaxvalOp, hlfir::MinvalOp, hlfir::MinlocOp,
- hlfir::MaxlocOp>();
- target.markUnknownOpDynamicallyLegal(
- [](mlir::Operation *) { return true; });
- if (mlir::failed(
- mlir::applyFullConversion(module, target, std::move(patterns)))) {
+
+ // While conceptually this pass is performing dialect conversion, we use
+ // pattern rewrites here instead of dialect conversion because this pass
+ // looses array bounds from some of the expressions e.g.
+ // !hlfir.expr<2xi32> -> !hlfir.expr<?xi32>
+ // MLIR thinks this is a different type so dialect conversion fails.
+ // Pattern rewriting only requires that the resulting IR is still valid
+ mlir::GreedyRewriteConfig config;
+ // Prevent the pattern driver from merging blocks
+ config.enableRegionSimplification = false;
+
+ if (mlir::failed(mlir::applyPatternsAndFoldGreedily(
+ module, std::move(patterns), config))) {
mlir::emitError(mlir::UnknownLoc::get(context),
"failure in HLFIR intrinsic lowering");
signalPassFailure();
diff --git a/flang/test/HLFIR/all-lowering.fir b/flang/test/HLFIR/all-lowering.fir
index dfd1ace947d68d..e83378eacf9c9f 100644
--- a/flang/test/HLFIR/all-lowering.fir
+++ b/flang/test/HLFIR/all-lowering.fir
@@ -34,6 +34,7 @@ func.func @_QPall2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?x!fir.logical<4>>>
// CHECK: %[[ARG2:.*]]: !fir.ref<i32>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_VAR:.*]]:2 = hlfir.declare %[[ARG2]]
// CHECK-DAG: %[[RES:.*]]:2 = hlfir.declare %[[ARG1]]
@@ -55,7 +56,6 @@ func.func @_QPall2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]#0
// CHECK: hlfir.destroy %[[EXPR]]
@@ -79,6 +79,7 @@ func.func @_QPall3(%arg0: !fir.ref<!fir.array<2x!fir.logical<4>>> {fir.bindc_nam
}
// CHECK-LABEL: func.func @_QPall3(
// CHECK: %[[ARG0:.*]]: !fir.ref<!fir.array<2x!fir.logical<4>>>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[RET_BOX:.*]] = fir.alloca !fir.box<!fir.heap<!fir.array<?x!fir.logical<4>>>>
// CHECK-DAG: %[[RET_ADDR:.*]] = fir.zero_bits !fir.heap<!fir.array<?x!fir.logical<4>>>
// CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
@@ -101,7 +102,6 @@ func.func @_QPall3(%arg0: !fir.ref<!fir.array<2x!fir.logical<4>>> {fir.bindc_nam
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]
// CHECK: hlfir.destroy %[[EXPR]]
@@ -125,6 +125,7 @@ func.func @_QPall4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?x!fir.logical<4>>>
// CHECK: %[[ARG2:.*]]: !fir.ref<!fir.box<!fir.ptr<i32>>>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_ARG:.*]]:2 = hlfir.declare %[[ARG2]]
// CHECK-DAG: %[[RES:.*]]:2 = hlfir.declare %[[ARG1]]
@@ -149,9 +150,8 @@ func.func @_QPall4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]
// CHECK: hlfir.destroy %[[EXPR]]
// CHECK-NEXT: return
-// CHECK-NEXT: }
\ No newline at end of file
+// CHECK-NEXT: }
diff --git a/flang/test/HLFIR/any-lowering.fir b/flang/test/HLFIR/any-lowering.fir
index ef8b8950293190..039146727d3f56 100644
--- a/flang/test/HLFIR/any-lowering.fir
+++ b/flang/test/HLFIR/any-lowering.fir
@@ -36,6 +36,7 @@ func.func @_QPany2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?x!fir.logical<4>>>
// CHECK: %[[ARG2:.*]]: !fir.ref<i32>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_VAR:.*]]:2 = hlfir.declare %[[ARG2]]
// CHECK-DAG: %[[RES:.*]]:2 = hlfir.declare %[[ARG1]]
@@ -57,7 +58,6 @@ func.func @_QPany2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]#0
// CHECK: hlfir.destroy %[[EXPR]]
@@ -82,6 +82,7 @@ func.func @_QPany3(%arg0: !fir.ref<!fir.array<2x!fir.logical<4>>> {fir.bindc_nam
}
// CHECK-LABEL: func.func @_QPany3(
// CHECK: %[[ARG0:.*]]: !fir.ref<!fir.array<2x!fir.logical<4>>>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[RET_BOX:.*]] = fir.alloca !fir.box<!fir.heap<!fir.array<?x!fir.logical<4>>>>
// CHECK-DAG: %[[RET_ADDR:.*]] = fir.zero_bits !fir.heap<!fir.array<?x!fir.logical<4>>>
// CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
@@ -104,7 +105,6 @@ func.func @_QPany3(%arg0: !fir.ref<!fir.array<2x!fir.logical<4>>> {fir.bindc_nam
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]
// CHECK: hlfir.destroy %[[EXPR]]
@@ -129,6 +129,7 @@ func.func @_QPany4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?x!fir.logical<4>>>
// CHECK: %[[ARG2:.*]]: !fir.ref<!fir.box<!fir.ptr<i32>>>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_ARG:.*]]:2 = hlfir.declare %[[ARG2]]
// CHECK-DAG: %[[RES:.*]]:2 = hlfir.declare %[[ARG1]]
@@ -153,7 +154,6 @@ func.func @_QPany4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_n
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?x!fir.logical<4>>>, i1) -> !hlfir.expr<?x!fir.logical<4>>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]
// CHECK: hlfir.destroy %[[EXPR]]
diff --git a/flang/test/HLFIR/count-lowering-default-int-kinds.fir b/flang/test/HLFIR/count-lowering-default-int-kinds.fir
index ea66c435e6a8a7..68bc7fdbaad876 100644
--- a/flang/test/HLFIR/count-lowering-default-int-kinds.fir
+++ b/flang/test/HLFIR/count-lowering-default-int-kinds.fir
@@ -2,9 +2,9 @@
// RUN: fir-opt %s -lower-hlfir-intrinsics | FileCheck %s
module attributes {fir.defaultkind = "a1c4d8i8l4r4", fir.kindmap = ""} {
- func.func @test_i8(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) {
+ func.func @test_i8(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) -> !hlfir.expr<?xi64> {
%4 = hlfir.count %arg0 dim %arg1 : (!fir.box<!fir.array<?x?x!fir.logical<4>>>, i64) -> !hlfir.expr<?xi64>
- return
+ return %4 : !hlfir.expr<?xi64>
}
}
// CHECK-LABEL: func.func @test_i8
@@ -12,9 +12,9 @@ module attributes {fir.defaultkind = "a1c4d8i8l4r4", fir.kindmap = ""} {
// CHECK: fir.call @_FortranACountDim(%{{.*}}, %{{.*}}, %{{.*}}, %[[KIND]], %{{.*}}, %{{.*}}) : (!fir.ref<!fir.box<none>>, !fir.box<none>, i32, i32, !fir.ref<i8>, i32) -> none
module attributes {fir.defaultkind = "a1c4d8i4l4r4", fir.kindmap = ""} {
- func.func @test_i4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) {
+ func.func @test_i4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) -> !hlfir.expr<?xi32> {
%4 = hlfir.count %arg0 dim %arg1 : (!fir.box<!fir.array<?x?x!fir.logical<4>>>, i64) -> !hlfir.expr<?xi32>
- return
+ return %4 : !hlfir.expr<?xi32>
}
}
// CHECK-LABEL: func.func @test_i4
@@ -22,9 +22,9 @@ module attributes {fir.defaultkind = "a1c4d8i4l4r4", fir.kindmap = ""} {
// CHECK: fir.call @_FortranACountDim(%{{.*}}, %{{.*}}, %{{.*}}, %[[KIND]], %{{.*}}, %{{.*}}) : (!fir.ref<!fir.box<none>>, !fir.box<none>, i32, i32, !fir.ref<i8>, i32) -> none
module attributes {fir.defaultkind = "a1c4d8i2l4r4", fir.kindmap = ""} {
- func.func @test_i2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) {
+ func.func @test_i2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) -> !hlfir.expr<?xi16> {
%4 = hlfir.count %arg0 dim %arg1 : (!fir.box<!fir.array<?x?x!fir.logical<4>>>, i64) -> !hlfir.expr<?xi16>
- return
+ return %4 : !hlfir.expr<?xi16>
}
}
// CHECK-LABEL: func.func @test_i2
@@ -32,9 +32,9 @@ module attributes {fir.defaultkind = "a1c4d8i2l4r4", fir.kindmap = ""} {
// CHECK: fir.call @_FortranACountDim(%{{.*}}, %{{.*}}, %{{.*}}, %[[KIND]], %{{.*}}, %{{.*}}) : (!fir.ref<!fir.box<none>>, !fir.box<none>, i32, i32, !fir.ref<i8>, i32) -> none
module attributes {fir.defaultkind = "a1c4d8i1l4r4", fir.kindmap = ""} {
- func.func @test_i1(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) {
+ func.func @test_i1(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc_name = "x"}, %arg1: i64) -> !hlfir.expr<?xi8> {
%4 = hlfir.count %arg0 dim %arg1 : (!fir.box<!fir.array<?x?x!fir.logical<4>>>, i64) -> !hlfir.expr<?xi8>
- return
+ return %4 : !hlfir.expr<?xi8>
}
}
// CHECK-LABEL: func.func @test_i1
diff --git a/flang/test/HLFIR/count-lowering.fir b/flang/test/HLFIR/count-lowering.fir
index da0f250dceef35..c3309724981a3f 100644
--- a/flang/test/HLFIR/count-lowering.fir
+++ b/flang/test/HLFIR/count-lowering.fir
@@ -34,6 +34,7 @@ func.func @_QPcount2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?xi32>
// CHECK: %[[ARG2:.*]]: !fir.ref<i32>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[KIND:.*]] = arith.constant 4 : i32
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_VAR:.*]]:2 = hlfir.declare %[[ARG2]]
@@ -56,7 +57,6 @@ func.func @_QPcount2(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?xi32>>, i1) -> !hlfir.expr<?xi32>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]#0
// CHECK: hlfir.destroy %[[EXPR]]
@@ -80,6 +80,7 @@ func.func @_QPcount3(%arg0: !fir.ref<!fir.array<2xi32>> {fir.bindc_name = "s"})
}
// CHECK-LABEL: func.func @_QPcount3(
// CHECK: %[[ARG0:.*]]: !fir.ref<!fir.array<2xi32>>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[RET_BOX:.*]] = fir.alloca !fir.box<!fir.heap<!fir.array<?xi32>>>
// CHECK-DAG: %[[KIND:.*]] = arith.constant 4 : i32
// CHECK-DAG: %[[RET_ADDR:.*]] = fir.zero_bits !fir.heap<!fir.array<?xi32>>
@@ -104,7 +105,6 @@ func.func @_QPcount3(%arg0: !fir.ref<!fir.array<2xi32>> {fir.bindc_name = "s"})
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?xi32>>, i1) -> !hlfir.expr<?xi32>
// CHECK: hlfir.assign %[[EXPR]] to %[[RES]]
// CHECK: hlfir.destroy %[[EXPR]]
@@ -133,6 +133,7 @@ func.func @_QPcount4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc
// CHECK: %[[ARG0:.*]]: !fir.box<!fir.array<?x?x!fir.logical<4>>>
// CHECK: %[[ARG1:.*]]: !fir.box<!fir.array<?xi32>
// CHECK: %[[ARG2:.*]]: !fir.ref<i32>
+// CHECK-DAG: %[[TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[MASK:.*]]:2 = hlfir.declare %[[ARG0]]
// CHECK-DAG: %[[DIM_VAR:.*]]:2 = hlfir.declare %[[ARG2]]
// CHECK-DAG: %[[RES:.*]]:2 = hlfir.declare %[[ARG1]]
@@ -155,7 +156,6 @@ func.func @_QPcount4(%arg0: !fir.box<!fir.array<?x?x!fir.logical<4>>> {fir.bindc
// CHECK-NEXT: %[[ADDR:.*]] = fir.box_addr %[[RET]]
// CHECK-NEXT: %[[SHIFT:.*]] = fir.shape_shift %[[BOX_DIMS]]#0, %[[BOX_DIMS]]#1
// CHECK-NEXT: %[[TMP:.*]]:2 = hlfir.declare %[[ADDR]](%[[SHIFT]]) {uniq_name = ".tmp.intrinsic_result"}
-// CHECK: %[[TRUE:.*]] = arith.constant true
// CHECK: %[[EXPR:.*]] = hlfir.as_expr %[[TMP]]#0 move %[[TRUE]] : (!fir.box<!fir.array<?xi64>>, i1) -> !hlfir.expr<?xi64>
// CHECK-NEXT: %[[OUT_SHAPE:.*]] = hlfir.shape_of %[[EXPR]]
// CHECK-NEXT: %[[OUT:.*]] = hlfir.elemental %[[OUT_SHAPE]] : (!fir.shape<1>) -> !hlfir.expr<?xi32>
diff --git a/flang/test/HLFIR/dot_product-lowering.fir b/flang/test/HLFIR/dot_product-lowering.fir
index e4f91eabfc0991..64d65665433f15 100644
--- a/flang/test/HLFIR/dot_product-lowering.fir
+++ b/flang/test/HLFIR/dot_product-lowering.fir
@@ -96,7 +96,6 @@ func.func @_QPdot_product4(%arg0: !fir.box<!fir.array<?x!fir.logical<4>>> {fir.b
// CHECK: %[[VAL_2:.*]] = fir.alloca !fir.logical<4>
// CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_0]] {uniq_name = "_QFdot_product2Elhs"} : (!fir.box<!fir.array<?x!fir.logical<4>>>) -> (!fir.box<!fir.array<?x!fir.logical<4>>>, !fir.box<!fir.array<?x!fir.logical<4>>>)
// CHECK: %[[VAL_4:.*]]:2 = hlfir.declare %[[VAL_1]] {uniq_name = "_QFdot_product2Erhs"} : (!fir.box<!fir.array<?x!fir.logical<4>>>) -> (!fir.box<!fir.array<?x!fir.logical<4>>>, !fir.box<!fir.array<?x!fir.logical<4>>>)
-// CHECK: %[[VAL_5:.*]] = fir.absent !fir.box<!fir.logical<4>>
// CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_3]]#1 : (!fir.box<!fir.array<?x!fir.logical<4>>>) -> !fir.box<none>
// CHECK: %[[VAL_10:.*]] = fir.convert %[[VAL_4]]#1 : (!fir.box<!fir.array<?x!fir.logical<4>>>) -> !fir.box<none>
// CHECK: %[[VAL_12:.*]] = fir.call @_FortranADotProductLogical(%[[VAL_9]], %[[VAL_10]], %{{.*}}, %{{.*}}) fastmath<contract> : (!fir.box<none>, !fir.box<none>, !fir.ref<i8>, i32) -> i1
diff --git a/flang/test/HLFIR/extents-of-shape-of.f90 b/flang/test/HLFIR/extents-of-shape-of.f90
index d807f8b70302b6..1168004597d191 100644
--- a/flang/test/HLFIR/extents-of-shape-of.f90
+++ b/flang/test/HLFIR/extents-of-shape-of.f90
@@ -31,18 +31,17 @@ elemental subroutine elem_sub(x)
! CHECK-HLFIR-NEXT: hlfir.destroy %[[MUL]]
! ...
+! CHECK-FIR-DAG: %[[C0:.*]] = arith.constant 0 : index
+! CHECK-FIR-DAG: %[[C1:.*]] = arith.constant 1 : index
+! CHECK-FIR-DAG: %[[C2:.*]] = arith.constant 2 : index
! CHECK-FIR: fir.call @_FortranAMatmul
! CHECK-FIR-NEXT: %[[MUL:.*]] = fir.load %[[MUL_BOX:.*]] : !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xf32>>>>
-! CHECK-FIR-NEXT: %[[C0:.*]] = arith.constant 0 : index
! CHECK-FIR-NEXT: %[[DIMS0:.*]]:3 = fir.box_dims %[[MUL]], %[[C0]]
-! CHECK-FIR-NEXT: %[[C1:.*]] = arith.constant 1 : index
! CHECK-FIR-NEXT: %[[DIMS1:.*]]:3 = fir.box_dims %[[MUL]], %[[C1]]
! ...
! CHECK-FIR: ...
[truncated]
|
matthias-springer
approved these changes
Feb 29, 2024
There was a problem hiding this comment.
Thanks!
I'm thinking of adding additional options to GreedyRewriteConfig to deactivate folding and/or CSE'ing/moving of constants. That should make changes like this one easier in the future and also generally gives more control over what's happening during a greedy pattern rewrite.
flang/test/HLFIR/minloc-lowering.fir
Outdated
| // CHECK-NEXT: %[[V0:.*]] = fir.alloca !fir.box<!fir.heap<!fir.array<?xi32>>> | ||
| // CHECK-DAG: %[[TRUE:.*]] = arith.constant true | ||
| // CHECK-DAG: %[[FALSE:.*]] = arith.constant false | ||
| // CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In #83253 @matthias-springer pointed out that LowerHLFIRIntrinsics.cpp should not be using rewrite patterns with the dialect conversion driver.
The intention of this pass is to lower HLFIR intrinsic operations into FIR so it conceptually fits dialect conversion. However, dialect conversion is much stricter about changing types when replacing operations. This pass sometimes looses track of array bounds, resulting in replacements with operations with different but compatible types (expressions of the same rank and element types but with or without compile time known array bounds). This is difficult to accommodate with the dialect conversion driver and so I have changed to use the greedy pattern rewriter.
There is a lot of test churn because the greedy pattern rewriter also performs canonicalization.