[flang] Fixing code reordering in alloca hoist pass. #95610
Closed
Conversation
This file contains hidden or 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
llvmbot added flang 
Member
| @llvm/pr-subscribers-flang-codegen @llvm/pr-subscribers-flang-fir-hlfir Author: Vijay Kandiah (VijayKandiah) ChangesThis change avoids reordering allocas that are already in the proper block which my previous change was doing, and thus fixes the remaining lit test failures. Patch is 20.68 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/95610.diff 7 Files Affected:
diff --git a/flang/lib/Optimizer/CodeGen/CodeGen.cpp b/flang/lib/Optimizer/CodeGen/CodeGen.cpp index 4448224024f20..91ba28220d1d4 100644 --- a/flang/lib/Optimizer/CodeGen/CodeGen.cpp +++ b/flang/lib/Optimizer/CodeGen/CodeGen.cpp @@ -255,8 +255,10 @@ struct AllocaOpConversion : public fir::FIROpConversion<fir::AllocaOp> { mlir::Region *parentRegion = rewriter.getInsertionBlock()->getParent(); mlir::Block *insertBlock = getBlockForAllocaInsert(parentOp, parentRegion); - size.getDefiningOp()->moveAfter(insertBlock, insertBlock->begin()); - rewriter.setInsertionPointAfter(size.getDefiningOp()); + if (rewriter.getInsertionBlock() != insertBlock) { + size.getDefiningOp()->moveAfter(insertBlock, insertBlock->begin()); + rewriter.setInsertionPointAfter(size.getDefiningOp()); + } } // NOTE: we used to pass alloc->getAttrs() in the builder for non opaque @@ -3705,7 +3707,8 @@ class FIRToLLVMLowering signalPassFailure(); } - // Run pass to add comdats to functions that have weak linkage on relevant platforms + // Run pass to add comdats to functions that have weak linkage on relevant + // platforms if (fir::getTargetTriple(mod).supportsCOMDAT()) { mlir::OpPassManager comdatPM("builtin.module"); comdatPM.addPass(mlir::LLVM::createLLVMAddComdats()); diff --git a/flang/test/Fir/alloc.fir b/flang/test/Fir/alloc.fir index e00fc9d6649c4..cf77fcf8c6988 100644 --- a/flang/test/Fir/alloc.fir +++ b/flang/test/Fir/alloc.fir @@ -334,10 +334,10 @@ func.func @allocmem_array_with_holes_dynchar(%arg0: index, %arg1: index) -> !fir } // CHECK-LABEL: define void @alloca_unlimited_polymorphic_box -// CHECK: %[[VAL_1:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]], ptr, [1 x i64] }, i64 1 // CHECK: %[[VAL_0:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, ptr, [1 x i64] }, i64 1 -// CHECK: %[[VAL_3:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]], ptr, [1 x i64] }, i64 1 +// CHECK: %[[VAL_1:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]], ptr, [1 x i64] }, i64 1 // CHECK: %[[VAL_2:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, ptr, [1 x i64] }, i64 1 +// CHECK: %[[VAL_3:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]], ptr, [1 x i64] }, i64 1 func.func @alloca_unlimited_polymorphic_box() { %0 = fir.alloca !fir.class<none> diff --git a/flang/test/Fir/boxproc.fir b/flang/test/Fir/boxproc.fir index 834017bff71aa..11d11f24d547e 100644 --- a/flang/test/Fir/boxproc.fir +++ b/flang/test/Fir/boxproc.fir @@ -1,12 +1,12 @@ // RUN: tco %s | FileCheck %s // CHECK-LABEL: define void @_QPtest_proc_dummy() -// CHECK: %[[VAL_3:.*]] = alloca [32 x i8], i64 1, align 1 -// CHECK: %[[VAL_1:.*]] = alloca { ptr }, i64 1, align 8 // CHECK: %[[VAL_0:.*]] = alloca i32, i64 1, align 4 +// CHECK: %[[VAL_1:.*]] = alloca { ptr }, i64 1, align 8 // CHECK: %[[VAL_2:.*]] = getelementptr { ptr }, ptr %[[VAL_1]], i32 0, i32 0 // CHECK: store ptr %[[VAL_0]], ptr %[[VAL_2]], align 8 // CHECK: store i32 1, ptr %[[VAL_0]], align 4 +// CHECK: %[[VAL_3:.*]] = alloca [32 x i8], i64 1, align 1 // CHECK: call void @llvm.init.trampoline(ptr %[[VAL_3]], ptr @_QFtest_proc_dummyPtest_proc_dummy_a, ptr %[[VAL_1]]) // CHECK: %[[VAL_6:.*]] = call ptr @llvm.adjust.trampoline(ptr %[[VAL_3]]) // CHECK: call void @_QPtest_proc_dummy_other(ptr %[[VAL_6]]) @@ -62,9 +62,9 @@ func.func @_QPtest_proc_dummy_other(%arg0: !fir.boxproc<() -> ()>) { // CHECK-LABEL: define void @_QPtest_proc_dummy_char() // CHECK: %[[VAL_20:.*]] = alloca [32 x i8], i64 1, align 1 -// CHECK: %[[VAL_2:.*]] = alloca { { ptr, i64 } }, i64 1, align 8 -// CHECK: %[[VAL_1:.*]] = alloca [10 x i8], i64 1, align 1 // CHECK: %[[VAL_0:.*]] = alloca [40 x i8], i64 1, align 1 +// CHECK: %[[VAL_1:.*]] = alloca [10 x i8], i64 1, align 1 +// CHECK: %[[VAL_2:.*]] = alloca { { ptr, i64 } }, i64 1, align 8 // CHECK: %[[VAL_3:.*]] = getelementptr { { ptr, i64 } }, ptr %[[VAL_2]], i32 0, i32 0 // CHECK: %[[VAL_5:.*]] = insertvalue { ptr, i64 } undef, ptr %[[VAL_1]], 0 // CHECK: %[[VAL_6:.*]] = insertvalue { ptr, i64 } %[[VAL_5]], i64 10, 1 diff --git a/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir index 45ff89bc40943..6c388f93dcbc5 100644 --- a/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir +++ b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir @@ -283,52 +283,52 @@ func.func @_QPomp_target_data() { // CHECK-LABEL: llvm.func @_QPomp_target_data() { // CHECK: %[[VAL_0:.*]] = llvm.mlir.constant(1024 : index) : i64 // CHECK: %[[VAL_1:.*]] = llvm.mlir.constant(1 : i64) : i64 -// CHECK: %[[VAL_2:.*]] = llvm.alloca %[[VAL_1]] x !llvm.array<1024 x i32> {bindc_name = "d"} : (i64) -> !llvm.ptr -// CHECK: %[[VAL_3:.*]] = llvm.mlir.constant(1 : i64) : i64 -// CHECK: %[[VAL_4:.*]] = llvm.alloca %[[VAL_3]] x !llvm.array<1024 x i32> {bindc_name = "c"} : (i64) -> !llvm.ptr -// CHECK: %[[VAL_5:.*]] = llvm.mlir.constant(1 : i64) : i64 -// CHECK: %[[VAL_6:.*]] = llvm.alloca %[[VAL_5]] x !llvm.array<1024 x i32> {bindc_name = "b"} : (i64) -> !llvm.ptr +// CHECK: %[[VAL_2:.*]] = llvm.alloca %[[VAL_1]] x !llvm.array<1024 x i32> {bindc_name = "a"} : (i64) -> !llvm.ptr +// CHECK: %[[VAL_3:.*]] = llvm.mlir.constant(1024 : index) : i64 +// CHECK: %[[VAL_4:.*]] = llvm.mlir.constant(1 : i64) : i64 +// CHECK: %[[VAL_5:.*]] = llvm.alloca %[[VAL_4]] x !llvm.array<1024 x i32> {bindc_name = "b"} : (i64) -> !llvm.ptr +// CHECK: %[[VAL_6:.*]] = llvm.mlir.constant(1024 : index) : i64 // CHECK: %[[VAL_7:.*]] = llvm.mlir.constant(1 : i64) : i64 -// CHECK: %[[VAL_8:.*]] = llvm.alloca %[[VAL_7]] x !llvm.array<1024 x i32> {bindc_name = "a"} : (i64) -> !llvm.ptr +// CHECK: %[[VAL_8:.*]] = llvm.alloca %[[VAL_7]] x !llvm.array<1024 x i32> {bindc_name = "c"} : (i64) -> !llvm.ptr // CHECK: %[[VAL_9:.*]] = llvm.mlir.constant(1024 : index) : i64 -// CHECK: %[[VAL_10:.*]] = llvm.mlir.constant(1024 : index) : i64 -// CHECK: %[[VAL_11:.*]] = llvm.mlir.constant(1024 : index) : i64 +// CHECK: %[[VAL_10:.*]] = llvm.mlir.constant(1 : i64) : i64 +// CHECK: %[[VAL_11:.*]] = llvm.alloca %[[VAL_10]] x !llvm.array<1024 x i32> {bindc_name = "d"} : (i64) -> !llvm.ptr // CHECK: %[[VAL_12:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_13:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_14:.*]] = llvm.mlir.constant(1023 : index) : i64 // CHECK: %[[VAL_15:.*]] = omp.map.bounds lower_bound(%[[VAL_13]] : i64) upper_bound(%[[VAL_14]] : i64) extent(%[[VAL_0]] : i64) stride(%[[VAL_12]] : i64) start_idx(%[[VAL_12]] : i64) -// CHECK: %[[VAL_16:.*]] = omp.map.info var_ptr(%[[VAL_8]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(to) capture(ByRef) bounds(%[[VAL_15]]) -> !llvm.ptr {name = "a"} +// CHECK: %[[VAL_16:.*]] = omp.map.info var_ptr(%[[VAL_2]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(to) capture(ByRef) bounds(%[[VAL_15]]) -> !llvm.ptr {name = "a"} // CHECK: %[[VAL_17:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_18:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_19:.*]] = llvm.mlir.constant(1023 : index) : i64 -// CHECK: %[[VAL_20:.*]] = omp.map.bounds lower_bound(%[[VAL_18]] : i64) upper_bound(%[[VAL_19]] : i64) extent(%[[VAL_9]] : i64) stride(%[[VAL_17]] : i64) start_idx(%[[VAL_17]] : i64) -// CHECK: %[[VAL_21:.*]] = omp.map.info var_ptr(%[[VAL_6]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(to) capture(ByRef) bounds(%[[VAL_20]]) -> !llvm.ptr {name = "b"} +// CHECK: %[[VAL_20:.*]] = omp.map.bounds lower_bound(%[[VAL_18]] : i64) upper_bound(%[[VAL_19]] : i64) extent(%[[VAL_3]] : i64) stride(%[[VAL_17]] : i64) start_idx(%[[VAL_17]] : i64) +// CHECK: %[[VAL_21:.*]] = omp.map.info var_ptr(%[[VAL_5]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(to) capture(ByRef) bounds(%[[VAL_20]]) -> !llvm.ptr {name = "b"} // CHECK: %[[VAL_22:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_23:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_24:.*]] = llvm.mlir.constant(1023 : index) : i64 -// CHECK: %[[VAL_25:.*]] = omp.map.bounds lower_bound(%[[VAL_23]] : i64) upper_bound(%[[VAL_24]] : i64) extent(%[[VAL_10]] : i64) stride(%[[VAL_22]] : i64) start_idx(%[[VAL_22]] : i64) -// CHECK: %[[VAL_26:.*]] = omp.map.info var_ptr(%[[VAL_4]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(always, exit_release_or_enter_alloc) capture(ByRef) bounds(%[[VAL_25]]) -> !llvm.ptr {name = "c"} +// CHECK: %[[VAL_25:.*]] = omp.map.bounds lower_bound(%[[VAL_23]] : i64) upper_bound(%[[VAL_24]] : i64) extent(%[[VAL_6]] : i64) stride(%[[VAL_22]] : i64) start_idx(%[[VAL_22]] : i64) +// CHECK: %[[VAL_26:.*]] = omp.map.info var_ptr(%[[VAL_8]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(always, exit_release_or_enter_alloc) capture(ByRef) bounds(%[[VAL_25]]) -> !llvm.ptr {name = "c"} // CHECK: omp.target_enter_data map_entries(%[[VAL_16]], %[[VAL_21]], %[[VAL_26]] : !llvm.ptr, !llvm.ptr, !llvm.ptr) // CHECK: %[[VAL_27:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_28:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_29:.*]] = llvm.mlir.constant(1023 : index) : i64 // CHECK: %[[VAL_30:.*]] = omp.map.bounds lower_bound(%[[VAL_28]] : i64) upper_bound(%[[VAL_29]] : i64) extent(%[[VAL_0]] : i64) stride(%[[VAL_27]] : i64) start_idx(%[[VAL_27]] : i64) -// CHECK: %[[VAL_31:.*]] = omp.map.info var_ptr(%[[VAL_8]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(from) capture(ByRef) bounds(%[[VAL_30]]) -> !llvm.ptr {name = "a"} +// CHECK: %[[VAL_31:.*]] = omp.map.info var_ptr(%[[VAL_2]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(from) capture(ByRef) bounds(%[[VAL_30]]) -> !llvm.ptr {name = "a"} // CHECK: %[[VAL_32:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_33:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_34:.*]] = llvm.mlir.constant(1023 : index) : i64 -// CHECK: %[[VAL_35:.*]] = omp.map.bounds lower_bound(%[[VAL_33]] : i64) upper_bound(%[[VAL_34]] : i64) extent(%[[VAL_9]] : i64) stride(%[[VAL_32]] : i64) start_idx(%[[VAL_32]] : i64) -// CHECK: %[[VAL_36:.*]] = omp.map.info var_ptr(%[[VAL_6]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(from) capture(ByRef) bounds(%[[VAL_35]]) -> !llvm.ptr {name = "b"} +// CHECK: %[[VAL_35:.*]] = omp.map.bounds lower_bound(%[[VAL_33]] : i64) upper_bound(%[[VAL_34]] : i64) extent(%[[VAL_3]] : i64) stride(%[[VAL_32]] : i64) start_idx(%[[VAL_32]] : i64) +// CHECK: %[[VAL_36:.*]] = omp.map.info var_ptr(%[[VAL_5]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(from) capture(ByRef) bounds(%[[VAL_35]]) -> !llvm.ptr {name = "b"} // CHECK: %[[VAL_37:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_38:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_39:.*]] = llvm.mlir.constant(1023 : index) : i64 -// CHECK: %[[VAL_40:.*]] = omp.map.bounds lower_bound(%[[VAL_38]] : i64) upper_bound(%[[VAL_39]] : i64) extent(%[[VAL_10]] : i64) stride(%[[VAL_37]] : i64) start_idx(%[[VAL_37]] : i64) -// CHECK: %[[VAL_41:.*]] = omp.map.info var_ptr(%[[VAL_4]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) bounds(%[[VAL_40]]) -> !llvm.ptr {name = "c"} +// CHECK: %[[VAL_40:.*]] = omp.map.bounds lower_bound(%[[VAL_38]] : i64) upper_bound(%[[VAL_39]] : i64) extent(%[[VAL_6]] : i64) stride(%[[VAL_37]] : i64) start_idx(%[[VAL_37]] : i64) +// CHECK: %[[VAL_41:.*]] = omp.map.info var_ptr(%[[VAL_8]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) bounds(%[[VAL_40]]) -> !llvm.ptr {name = "c"} // CHECK: %[[VAL_42:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_43:.*]] = llvm.mlir.constant(0 : index) : i64 // CHECK: %[[VAL_44:.*]] = llvm.mlir.constant(1023 : index) : i64 -// CHECK: %[[VAL_45:.*]] = omp.map.bounds lower_bound(%[[VAL_43]] : i64) upper_bound(%[[VAL_44]] : i64) extent(%[[VAL_11]] : i64) stride(%[[VAL_42]] : i64) start_idx(%[[VAL_42]] : i64) -// CHECK: %[[VAL_46:.*]] = omp.map.info var_ptr(%[[VAL_2]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(always, delete) capture(ByRef) bounds(%[[VAL_45]]) -> !llvm.ptr {name = "d"} +// CHECK: %[[VAL_45:.*]] = omp.map.bounds lower_bound(%[[VAL_43]] : i64) upper_bound(%[[VAL_44]] : i64) extent(%[[VAL_9]] : i64) stride(%[[VAL_42]] : i64) start_idx(%[[VAL_42]] : i64) +// CHECK: %[[VAL_46:.*]] = omp.map.info var_ptr(%[[VAL_11]] : !llvm.ptr, !llvm.array<1024 x i32>) map_clauses(always, delete) capture(ByRef) bounds(%[[VAL_45]]) -> !llvm.ptr {name = "d"} // CHECK: omp.target_exit_data map_entries(%[[VAL_31]], %[[VAL_36]], %[[VAL_41]], %[[VAL_46]] : !llvm.ptr, !llvm.ptr, !llvm.ptr, !llvm.ptr) // CHECK: llvm.return // CHECK: } @@ -374,9 +374,9 @@ func.func @_QPopenmp_target_data_region() { // CHECK-LABEL: llvm.func @_QPopenmp_target_data_region() { // CHECK: %[[VAL_0:.*]] = llvm.mlir.constant(1 : i64) : i64 +// CHECK: %[[VAL_1:.*]] = llvm.alloca %[[VAL_0]] x !llvm.array<1024 x i32> {bindc_name = "a"} : (i64) -> !llvm.ptr // CHECK: %[[VAL_2:.*]] = llvm.mlir.constant(1 : i64) : i64 // CHECK: %[[VAL_3:.*]] = llvm.alloca %[[VAL_2]] x i32 {bindc_name = "i"} : (i64) -> !llvm.ptr -// CHECK: %[[VAL_1:.*]] = llvm.alloca %[[VAL_0]] x !llvm.array<1024 x i32> {bindc_name = "a"} : (i64) -> !llvm.ptr // CHECK: %[[VAL_MAX:.*]] = llvm.mlir.constant(1024 : index) : i64 // CHECK: %[[VAL_ONE:.*]] = llvm.mlir.constant(1 : index) : i64 // CHECK: %[[VAL_ZERO:.*]] = llvm.mlir.constant(0 : index) : i64 @@ -675,9 +675,10 @@ func.func @_QPsb() { } // CHECK: llvm.func @_QPsb() { +// CHECK: %[[ONE:.*]] = llvm.mlir.constant(1 : i32) : i32 +// CHECK: %[[TWO:.*]] = llvm.mlir.constant(1 : i64) : i64 // CHECK: %[[SIZE:.*]] = llvm.mlir.constant(1 : i64) : i64 // CHECK: %[[LI_REF:.*]] = llvm.alloca %[[SIZE]] x i32 {bindc_name = "li"} : (i64) -> !llvm.ptr -// CHECK: %[[ONE:.*]] = llvm.mlir.constant(1 : i32) : i32 // CHECK: omp.sections { // CHECK: omp.section { // CHECK: llvm.br ^[[BB_ENTRY:.*]]({{.*}}) @@ -715,7 +716,7 @@ func.func @_QPsb() { // CHECK: } // CHECK-LABEL: @_QPsimple_reduction // CHECK-SAME: %[[ARRAY_REF:.*]]: !llvm.ptr -// CHECK: %[[RED_ACCUMULATOR:.*]] = llvm.alloca %1 x i32 {bindc_name = "x"} : (i64) -> !llvm.ptr +// CHECK: %[[RED_ACCUMULATOR:.*]] = llvm.alloca %2 x i32 {bindc_name = "x"} : (i64) -> !llvm.ptr // CHECK: omp.parallel { // CHECK: omp.wsloop reduction(@[[EQV_REDUCTION]] %[[RED_ACCUMULATOR]] -> %[[PRV:.+]] : !llvm.ptr) { // CHECK-NEXT: omp.loop_nest @@ -797,7 +798,6 @@ func.func @_QPs(%arg0: !fir.ref<!fir.complex<4>> {fir.bindc_name = "x"}) { // Test if llvm.alloca is properly inserted in the omp section -//CHECK: %[[CONST0:.*]] = llvm.mlir.constant(1 : i64) : i64 //CHECK: %[[CONST:.*]] = llvm.mlir.constant(1 : i64) : i64 //CHECK: %[[ALLOCA:.*]] = llvm.alloca %[[CONST]] x !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)> {bindc_name = "iattr"} : (i64) -> !llvm.ptr //CHECK: omp.parallel { @@ -908,9 +908,9 @@ omp.critical.declare @help hint(contended) // CHECK: llvm.func @omp_critical_() { func.func @omp_critical_() { -// CHECK: %[[Y_REF:.*]] = llvm.alloca %{{.*}} x i32 {bindc_name = "y"} : (i64) -> !llvm.ptr - %0 = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFomp_criticalEx"} // CHECK: %[[X_REF:.*]] = llvm.alloca %{{.*}} x i32 {bindc_name = "x"} : (i64) -> !llvm.ptr + %0 = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFomp_criticalEx"} +// CHECK: %[[Y_REF:.*]] = llvm.alloca %{{.*}} x i32 {bindc_name = "y"} : (i64) -> !llvm.ptr %1 = fir.alloca i32 {bindc_name = "y", uniq_name = "_QFomp_criticalEy"} // CHECK: omp.critical(@help) omp.critical(@help) { diff --git a/flang/test/Fir/convert-to-llvm.fir b/flang/test/Fir/convert-to-llvm.fir index d7059671d3a88..217bfc120e572 100644 --- a/flang/test/Fir/convert-to-llvm.fir +++ b/flang/test/Fir/convert-to-llvm.fir @@ -2015,17 +2015,17 @@ func.func private @_QPtest_dt_callee(%arg0: !fir.box<!fir.array<?xi32>>) // AMDGPU: %[[AA:.*]] = llvm.alloca %[[ALLOCA_SIZE]] x !llvm.struct<(ptr, i{{.*}}, i{{.*}}, i{{.*}}, i{{.*}}, i{{.*}}, i{{.*}}, array<1 x array<3 x i64>>)> {alignment = 8 : i64} : (i32) -> !llvm.ptr<5> // AMDGPU: %[[ALLOCA:.*]] = llvm.addrspacecast %[[AA]] : !llvm.ptr<5> to !llvm.ptr // CHECK: %[[C20:.*]] = llvm.mlir.constant(20 : index) : i64 -// CHECK: %[[ALLOCA_SIZE_X:.*]] = llvm.mlir.constant(1 : i64) : i64 -// GENERIC: %[[X:.*]] = llvm.alloca %[[ALLOCA_SIZE_X]] x !llvm.array<20 x struct<"_QFtest_dt_sliceTt", (i32, i32)>> {bindc_name = "x"} : (i64) -> !llvm.ptr -// AMDGPU: %[[AC:.*]] = llvm.alloca %[[ALLOCA_SIZE_X]] x !llvm.array<20 x struct<"_QFtest_dt_sliceTt", (i32, i32)>> {bindc_name = "x"} : (i64) -> !llvm.ptr<5> -// AMDGPU: %[[X:.*]] = llvm.addrspacecast %[[AC]] : !llvm.ptr<5> to !llvm.ptr +// CHECK: %[[C1:.*]] = llvm.mlir.constant(1 : i64) : i64 +// CHECK: %[[C10:.*]] = llvm.mlir.constant(10 : i64) : i64 +// CHECK: %[[C2:.*]] = llvm.mlir.constant(2 : i64) : i64 // CHECK: %[[ALLOCA_SIZE_V:.*]] = llvm.mlir.constant(1 : i64) : i64 // GENERIC: %[[V:.*]] = llvm.alloca %[[ALLOCA_SIZE_V]] x i32 {bindc_name = "v"} : (i64) -> !llvm.ptr // AMDGPU: %[[AB:.*]] = llvm.alloca %[[ALLOCA_SIZE_V]] x i32 {bindc_name = "v"} : (i64) -> !llvm.ptr<5> // AMDGPU: %[[V:.*]] = llvm.addrspacecast %[[AB]] : !llvm.ptr<5> to !llvm.ptr -// CHECK: %[[C1:.*]] = llvm.mlir.constant(1 : i64) : i64 -// CHECK: %[[C10:.*]] = llvm.mlir.constant(10 : i64) : i64 -// CHECK: %[[C2:.*]] = llvm.mlir.constant(2 : i64) : i64 +// CHECK: %[[ALLOCA_SIZE_X:.*]] = llvm.mlir.constant(1 : i64) : i64 +// GENERIC: %[[X:.*]] = llvm.alloca %[[ALLOCA_SIZE_X]] x !llvm.array<20 x struct<"_QFtest_dt_sliceTt", (i32, i32)>> {bindc_name = "x"} : (i64) -> !llvm.ptr +// AMDGPU: %[[AC:.*]] = llvm.alloca %[[ALLOCA_SIZE_X]] x !llvm.array<20 x struct<"_QFtest_dt_sliceTt", (i32, i32)>> {bindc_name = "x"} : (i64) -> !llvm.ptr<5> +// AMDGPU: %[[X:.*]] = llvm.addrspacecast %[[AC]] : !llvm.ptr<5> to !llvm.ptr // CHECK: %[[TYPE_CODE:.*]] = llvm.mlir.constant(9 : i32) : i32 // CHECK: %[[NULL:.*]] = llvm.mlir.zero : !llvm.ptr // CHECK: %[[GEP:.*]] = llvm.getelementptr %[[NULL]][1] diff --git a/flang/test/Integration/OpenMP/copyprivate.f90 b/flang/test/Integration/OpenMP/copyprivate.f90 index dd69ebdb881a1..d32319a18c28b 100644 --- a/flang/test/Integration/OpenMP/copyprivate.f90 +++ b/flang/test/Integration/OpenMP/copyprivate.f90 @@ -33,8 +33,8 @@ !CHECK-NEXT: } !CHECK-LABEL: define internal void @test_scalar_..omp_par({{.*}}) -!CHECK: %[[J:.*]] = alloca i32, i64 1 !CHECK: %[[I:.*]] = alloca i32, i64 1 +!CHECK: %[[J:.*]] = alloca i32, i64 1 !CHECK: %[[DID_IT:.*]] = alloca i32 !CHECK: store i32 0, ptr %[[DID_IT]] !CHECK: %[[THREAD_NUM1:.*]] = call i32 @__kmpc_global_thread_num(ptr @[[LOC:.*]]) diff --git a/flang/test/Transforms/debug-local-var-2.f90 b/flang/test/Transforms/debug-local-var-2.f90 index 79fe1bab6e35b..ce78bfd005056 100644 --- a/flang/test/Transforms/debug-local-var-2.f90 +++ b/flang/test/Transforms/debug-local-var-2.f90 @@ -6,12 +6,12 @@ ! This tests checks the debug information for local variables in llvm IR. ! BOTH-LABEL: define void @_QQmain -! BOTH-DAG: %[[AL16:.*]] = alloca double -! BOTH-DAG: %[[AL15:.*]] = alloca float -! BOTH-DAG: ... [truncated] |
vzakhari approved these changes Jun 14, 2024
| getBlockForAllocaInsert(parentOp, parentRegion); | ||
| size.getDefiningOp()->moveAfter(insertBlock, insertBlock->begin()); | ||
| rewriter.setInsertionPointAfter(size.getDefiningOp()); | ||
| if (rewriter.getInsertionBlock() != insertBlock) { |
Contributor
There was a problem hiding this comment.
This seems okay, but I think there might be cases where we will leave the allocas inside stacksave/stackrestore region (which the initial commit tried to avoid). I am okay with this change to resolve the LIT tests failures, but we may need to revisit this (and fix the tests, if needed).
Contributor Author
There was a problem hiding this comment.
I agree. We would need to revisit to move allocas outside stacksave/restore regions + fix lit tests as necessary.
Contributor Author
There was a problem hiding this comment.
On second thought, I think we would prefer to fix the remaining tests (5 powerpc tests) instead of making another codegen change. I'll go fix those and create a separate merge request to get rid of those lit test failures.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
3 participants
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.
This change avoids reordering allocas that are already in the proper block which my previous change was doing, and thus fixes the remaining lit test failures.