[llvm-branch-commits] [mlir] [MLIR][OpenMP] Support cancel taskgroup inside of taskloop (PR #174815)
via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Wed Jan 7 09:47:58 PST 2026
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-mlir-openmp
Author: Tom Eccles (tblah)
<details>
<summary>Changes</summary>
Implementation follows exactly what is done for omp.wsloop and omp.task. See #<!-- -->137841.
The change to the operation verifier is to allow a taskgroup cancellation point inside of a taskloop. This was already allowed for omp.cancel.
---
Patch is 50.95 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/174815.diff
4 Files Affected:
- (modified) mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp (+2-1)
- (modified) mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp (+12-20)
- (added) mlir/test/Target/LLVMIR/openmp-taskloop-cancel.mlir (+419)
- (added) mlir/test/Target/LLVMIR/openmp-taskloop-cancellation-point.mlir (+280)
``````````diff
diff --git a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
index 103295d136dbb..01dd510162a49 100644
--- a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
+++ b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
@@ -4129,7 +4129,8 @@ LogicalResult CancellationPointOp::verify() {
<< "inside a sections region";
}
if ((cct == ClauseCancellationConstructType::Taskgroup) &&
- !mlir::isa<omp::TaskOp>(structuralParent)) {
+ (!mlir::isa<omp::TaskOp>(structuralParent) &&
+ !mlir::isa<omp::TaskloopOp>(structuralParent->getParentOp()))) {
return emitOpError() << "cancellation point taskgroup must appear "
<< "inside a task region";
}
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index 43c503757ddf5..c64fbfe890e9a 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -329,22 +329,6 @@ static LogicalResult checkImplementationStatus(Operation &op) {
if (op.getBare())
result = todo("ompx_bare");
};
- auto checkCancelDirective = [&todo](auto op, LogicalResult &result) {
- omp::ClauseCancellationConstructType cancelledDirective =
- op.getCancelDirective();
- // Cancelling a taskloop is not yet supported because we don't yet have LLVM
- // IR conversion for taskloop
- if (cancelledDirective == omp::ClauseCancellationConstructType::Taskgroup) {
- Operation *parent = op->getParentOp();
- while (parent) {
- if (parent->getDialect() == op->getDialect())
- break;
- parent = parent->getParentOp();
- }
- if (isa_and_nonnull<omp::TaskloopOp>(parent))
- result = todo("cancel directive inside of taskloop");
- }
- };
auto checkCollapse = [&todo](auto op, LogicalResult &result) {
if (op.getCollapseNumLoops() > 1)
result = todo("collapse");
@@ -399,10 +383,6 @@ static LogicalResult checkImplementationStatus(Operation &op) {
LogicalResult result = success();
llvm::TypeSwitch<Operation &>(op)
- .Case([&](omp::CancelOp op) { checkCancelDirective(op, result); })
- .Case([&](omp::CancellationPointOp op) {
- checkCancelDirective(op, result);
- })
.Case([&](omp::DistributeOp op) {
checkAllocate(op, result);
checkOrder(op, result);
@@ -2773,6 +2753,16 @@ convertOmpTaskloopOp(Operation &opInst, llvm::IRBuilderBase &builder,
if (taskStructMgr.getStructPtr())
taskDupOrNull = taskDupCB;
+ llvm::OpenMPIRBuilder &ompBuilder = *moduleTranslation.getOpenMPBuilder();
+ SmallVector<llvm::BranchInst *> cancelTerminators;
+ // The directive to match here is OMPD_taskgroup because it is the
+ // taskgroup which is canceled. This is handled here because it is the
+ // task's cleanup block which should be branched to. It doesn't depend upon
+ // nogroup because even in that case the taskloop might still be inside an
+ // explicit taskgroup.
+ pushCancelFinalizationCB(cancelTerminators, builder, ompBuilder, taskloopOp,
+ llvm::omp::Directive::OMPD_taskgroup);
+
llvm::OpenMPIRBuilder::LocationDescription ompLoc(builder);
llvm::OpenMPIRBuilder::InsertPointOrErrorTy afterIP =
moduleTranslation.getOpenMPBuilder()->createTaskloop(
@@ -2789,6 +2779,8 @@ convertOmpTaskloopOp(Operation &opInst, llvm::IRBuilderBase &builder,
if (failed(handleError(afterIP, opInst)))
return failure();
+ popCancelFinalizationCB(cancelTerminators, ompBuilder, afterIP.get());
+
builder.restoreIP(*afterIP);
return success();
}
diff --git a/mlir/test/Target/LLVMIR/openmp-taskloop-cancel.mlir b/mlir/test/Target/LLVMIR/openmp-taskloop-cancel.mlir
new file mode 100644
index 0000000000000..7aa6398a7f113
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/openmp-taskloop-cancel.mlir
@@ -0,0 +1,419 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+omp.private {type = private} @_QFtestEi_private_i32 : i32
+omp.private {type = firstprivate} @_QFtestEarg_firstprivate_i32 : i32 init {
+^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr):
+ llvm.call @_init(%arg0, %arg1) : (!llvm.ptr, !llvm.ptr) -> ()
+ omp.yield(%arg1 : !llvm.ptr)
+} copy {
+^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr):
+ llvm.call @_copy(%arg0, %arg1) : (!llvm.ptr, !llvm.ptr) -> ()
+ omp.yield(%arg1 : !llvm.ptr)
+} dealloc {
+^bb0(%arg0: !llvm.ptr):
+ llvm.call @_dealloc(%arg0) : (!llvm.ptr) -> ()
+ omp.yield
+}
+
+// Test cancelling implicit taskgroup created by taskloop
+llvm.func @_QPtest(%arg0: !llvm.ptr {fir.bindc_name = "arg", llvm.noalias, llvm.nocapture}) {
+ %0 = llvm.mlir.constant(1 : i32) : i32
+ %1 = llvm.mlir.constant(100 : i32) : i32
+ %2 = llvm.mlir.constant(1 : i64) : i64
+ %3 = llvm.alloca %2 x i32 {bindc_name = "i"} : (i64) -> !llvm.ptr
+ omp.taskloop private(@_QFtestEarg_firstprivate_i32 %arg0 -> %arg1, @_QFtestEi_private_i32 %3 -> %arg2 : !llvm.ptr, !llvm.ptr) {
+ omp.loop_nest (%arg3) : i32 = (%0) to (%1) inclusive step (%0) {
+ llvm.store %arg3, %arg2 : i32, !llvm.ptr
+ llvm.call @_QPbefore(%arg1) : (!llvm.ptr) -> ()
+ omp.cancel cancellation_construct_type(taskgroup)
+ llvm.call @_QPafter(%arg1) : (!llvm.ptr) -> ()
+ omp.yield
+ }
+ }
+ llvm.return
+}
+// CHECK-LABEL: define void @_QPtest(
+// CHECK: %[[STRUCTARG:.*]] = alloca { i64, i64, i64, ptr }, align 8
+// CHECK: %[[VAL_0:.*]] = alloca i32, i64 1, align 4
+// CHECK: br label %[[VAL_1:.*]]
+// CHECK: entry: ; preds = %[[VAL_2:.*]]
+// CHECK: br label %[[VAL_3:.*]]
+// CHECK: omp.private.init: ; preds = %[[VAL_1]]
+// CHECK: %[[VAL_4:.*]] = tail call ptr @malloc(i64 ptrtoint (ptr getelementptr ({ i32 }, ptr null, i32 1) to i64))
+// CHECK: %[[VAL_5:.*]] = getelementptr { i32 }, ptr %[[VAL_4]], i32 0, i32 0
+// CHECK: call void @_init(ptr %[[VAL_6:.*]], ptr %[[VAL_5]])
+// CHECK: br label %[[VAL_7:.*]]
+// CHECK: omp.private.copy: ; preds = %[[VAL_3]]
+// CHECK: br label %[[VAL_8:.*]]
+// CHECK: omp.private.copy1: ; preds = %[[VAL_7]]
+// CHECK: call void @_copy(ptr %[[VAL_6]], ptr %[[VAL_5]])
+// CHECK: br label %[[VAL_9:.*]]
+// CHECK: omp.taskloop.start: ; preds = %[[VAL_8]]
+// CHECK: br label %[[VAL_10:.*]]
+// CHECK: codeRepl: ; preds = %[[VAL_9]]
+// CHECK: %[[VAL_11:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[STRUCTARG]], i32 0, i32 0
+// CHECK: store i64 1, ptr %[[VAL_11]], align 4
+// CHECK: %[[VAL_12:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[STRUCTARG]], i32 0, i32 1
+// CHECK: store i64 100, ptr %[[VAL_12]], align 4
+// CHECK: %[[VAL_13:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[STRUCTARG]], i32 0, i32 2
+// CHECK: store i64 1, ptr %[[VAL_13]], align 4
+// CHECK: %[[VAL_14:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[STRUCTARG]], i32 0, i32 3
+// CHECK: store ptr %[[VAL_4]], ptr %[[VAL_14]], align 8
+// CHECK: %[[VAL_15:.*]] = call i32 @__kmpc_global_thread_num(ptr @1)
+// CHECK: call void @__kmpc_taskgroup(ptr @1, i32 %[[VAL_15]])
+// CHECK: %[[VAL_16:.*]] = call ptr @__kmpc_omp_task_alloc(ptr @1, i32 %[[VAL_15]], i32 1, i64 40, i64 32, ptr @_QPtest..omp_par)
+// CHECK: %[[VAL_17:.*]] = load ptr, ptr %[[VAL_16]], align 8
+// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 1 %[[VAL_17]], ptr align 1 %[[STRUCTARG]], i64 32, i1 false)
+// CHECK: %[[VAL_18:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_17]], i32 0, i32 0
+// CHECK: %[[VAL_19:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_17]], i32 0, i32 1
+// CHECK: %[[VAL_20:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_17]], i32 0, i32 2
+// CHECK: %[[VAL_21:.*]] = load i64, ptr %[[VAL_20]], align 4
+// CHECK: call void @__kmpc_taskloop(ptr @1, i32 %[[VAL_15]], ptr %[[VAL_16]], i32 1, ptr %[[VAL_18]], ptr %[[VAL_19]], i64 %[[VAL_21]], i32 1, i32 0, i64 0, ptr @omp_taskloop_dup)
+// CHECK: call void @__kmpc_end_taskgroup(ptr @1, i32 %[[VAL_15]])
+// CHECK: br label %[[VAL_22:.*]]
+// CHECK: taskloop.exit: ; preds = %[[VAL_10]]
+// CHECK: ret void
+
+// CHECK-LABEL: define internal void @_QPtest..omp_par
+// CHECK: taskloop.alloca:
+// CHECK: %[[VAL_23:.*]] = load ptr, ptr %[[VAL_24:.*]], align 8
+// CHECK: %[[VAL_25:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_23]], i32 0, i32 0
+// CHECK: %[[VAL_26:.*]] = load i64, ptr %[[VAL_25]], align 4
+// CHECK: %[[VAL_27:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_23]], i32 0, i32 1
+// CHECK: %[[VAL_28:.*]] = load i64, ptr %[[VAL_27]], align 4
+// CHECK: %[[VAL_29:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_23]], i32 0, i32 2
+// CHECK: %[[VAL_30:.*]] = load i64, ptr %[[VAL_29]], align 4
+// CHECK: %[[VAL_31:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_23]], i32 0, i32 3
+// CHECK: %[[VAL_32:.*]] = load ptr, ptr %[[VAL_31]], align 8, !align !1
+// CHECK: %[[VAL_33:.*]] = alloca i32, align 4
+// CHECK: br label %[[VAL_34:.*]]
+// CHECK: taskloop.body: ; preds = %[[VAL_35:.*]]
+// CHECK: %[[VAL_36:.*]] = getelementptr { i32 }, ptr %[[VAL_32]], i32 0, i32 0
+// CHECK: br label %[[VAL_37:.*]]
+// CHECK: omp.taskloop.region: ; preds = %[[VAL_34]]
+// CHECK: br label %[[VAL_38:.*]]
+// CHECK: omp_loop.preheader: ; preds = %[[VAL_37]]
+// CHECK: %[[VAL_39:.*]] = sub i64 %[[VAL_28]], %[[VAL_26]]
+// CHECK: %[[VAL_40:.*]] = sdiv i64 %[[VAL_39]], 1
+// CHECK: %[[VAL_41:.*]] = add i64 %[[VAL_40]], 1
+// CHECK: %[[VAL_42:.*]] = trunc i64 %[[VAL_41]] to i32
+// CHECK: %[[VAL_43:.*]] = trunc i64 %[[VAL_26]] to i32
+// CHECK: br label %[[VAL_44:.*]]
+// CHECK: omp_loop.header: ; preds = %[[VAL_45:.*]], %[[VAL_38]]
+// CHECK: %[[VAL_46:.*]] = phi i32 [ 0, %[[VAL_38]] ], [ %[[VAL_47:.*]], %[[VAL_45]] ]
+// CHECK: br label %[[VAL_48:.*]]
+// CHECK: omp_loop.cond: ; preds = %[[VAL_44]]
+// CHECK: %[[VAL_49:.*]] = icmp ult i32 %[[VAL_46]], %[[VAL_42]]
+// CHECK: br i1 %[[VAL_49]], label %[[VAL_50:omp_loop.body]], label %[[VAL_51:omp_loop.exit]]
+// CHECK: omp_loop.exit: ; preds = %[[VAL_48]]
+// CHECK: br label %[[OMP_LOOP_AFTER:omp_loop.after]]
+// CHECK: omp_loop.after: ; preds = %[[VAL_51]]
+// CHECK: br label %[[CONT:omp.region.cont]]
+// CHECK: omp.region.cont: ; preds = %[[FINI:.fini]], %[[OMP_LOOP_AFTER]]
+// CHECK: call void @_dealloc(ptr %[[VAL_36]])
+// CHECK: tail call void @free(ptr %[[VAL_32]])
+// CHECK: br label %[[VAL_55:.*]]
+// CHECK: omp_loop.body: ; preds = %[[VAL_48]]
+// CHECK: %[[VAL_56:.*]] = mul i32 %[[VAL_46]], 1
+// CHECK: %[[VAL_57:.*]] = add i32 %[[VAL_56]], %[[VAL_43]]
+// CHECK: br label %[[LOOP_REGION:omp.loop_nest.region]]
+// CHECK: omp.loop_nest.region: ; preds = %[[VAL_50]]
+// CHECK: store i32 %[[VAL_57]], ptr %[[VAL_33]], align 4
+// CHECK: call void @_QPbefore(ptr %[[VAL_36]])
+// CHECK: %[[VAL_59:.*]] = call i32 @__kmpc_global_thread_num(ptr @1)
+// CHECK: %[[VAL_60:.*]] = call i32 @__kmpc_cancel(ptr @1, i32 %[[VAL_59]], i32 4)
+// CHECK: %[[VAL_61:.*]] = icmp eq i32 %[[VAL_60]], 0
+// CHECK: br i1 %[[VAL_61]], label %[[VAL_62:omp.loop_nest.region.split]], label %[[CNCL:omp.loop_nest.region.cncl]]
+// CHECK: omp.loop_nest.region.cncl: ; preds = %[[LOOP_REGION]]
+// CHECK: br label %[[FINI]]
+// CHECK: .fini: ; preds = %[[CNCL]]
+// CHECK: br label %[[CONT]]
+// CHECK: omp.loop_nest.region.split: ; preds = %[[LOOP_REGION]]
+// CHECK: call void @_QPafter(ptr %[[VAL_36]])
+// CHECK: br label %[[VAL_64:.*]]
+// CHECK: omp.region.cont2: ; preds = %[[VAL_62]]
+// CHECK: br label %[[VAL_45]]
+// CHECK: omp_loop.inc: ; preds = %[[VAL_64]]
+// CHECK: %[[VAL_47]] = add nuw i32 %[[VAL_46]], 1
+// CHECK: br label %[[VAL_44]]
+// CHECK: taskloop.exit.exitStub: ; preds = %[[CONT]]
+// CHECK: ret void
+
+// Tesk cancelling explicit taskgroup enclosing taskloop
+llvm.func @_QPtest2(%arg0: !llvm.ptr {fir.bindc_name = "arg", llvm.noalias, llvm.nocapture}) {
+ %0 = llvm.mlir.constant(1 : i32) : i32
+ %1 = llvm.mlir.constant(100 : i32) : i32
+ %2 = llvm.mlir.constant(1 : i64) : i64
+ %3 = llvm.alloca %2 x i32 {bindc_name = "i"} : (i64) -> !llvm.ptr
+ omp.taskgroup {
+ omp.taskloop nogroup private(@_QFtestEarg_firstprivate_i32 %arg0 -> %arg1, @_QFtestEi_private_i32 %3 -> %arg2 : !llvm.ptr, !llvm.ptr) {
+ omp.loop_nest (%arg3) : i32 = (%0) to (%1) inclusive step (%0) {
+ llvm.store %arg3, %arg2 : i32, !llvm.ptr
+ llvm.call @_QPbefore(%arg1) : (!llvm.ptr) -> ()
+ omp.cancel cancellation_construct_type(taskgroup)
+ llvm.call @_QPafter(%arg1) : (!llvm.ptr) -> ()
+ omp.yield
+ }
+ }
+ omp.terminator
+ }
+ llvm.return
+}
+// CHECK-LABEL: define void @_QPtest2(
+// CHECK: %[[VAL_65:.*]] = alloca { i64, i64, i64, ptr }, align 8
+// CHECK: %[[VAL_66:.*]] = alloca i32, i64 1, align 4
+// CHECK: br label %[[VAL_67:.*]]
+// CHECK: entry: ; preds = %[[VAL_68:.*]]
+// CHECK: %[[VAL_69:.*]] = call i32 @__kmpc_global_thread_num(ptr @1)
+// CHECK: call void @__kmpc_taskgroup(ptr @1, i32 %[[VAL_69]])
+// CHECK: br label %[[VAL_70:.*]]
+// CHECK: omp.taskgroup.region: ; preds = %[[VAL_67]]
+// CHECK: br label %[[VAL_71:.*]]
+// CHECK: omp.private.init: ; preds = %[[VAL_70]]
+// CHECK: %[[VAL_72:.*]] = tail call ptr @malloc(i64 ptrtoint (ptr getelementptr ({ i32 }, ptr null, i32 1) to i64))
+// CHECK: %[[VAL_73:.*]] = getelementptr { i32 }, ptr %[[VAL_72]], i32 0, i32 0
+// CHECK: call void @_init(ptr %[[VAL_74:.*]], ptr %[[VAL_73]])
+// CHECK: br label %[[VAL_75:.*]]
+// CHECK: omp.private.copy: ; preds = %[[VAL_71]]
+// CHECK: br label %[[VAL_76:.*]]
+// CHECK: omp.private.copy1: ; preds = %[[VAL_75]]
+// CHECK: call void @_copy(ptr %[[VAL_74]], ptr %[[VAL_73]])
+// CHECK: br label %[[VAL_77:.*]]
+// CHECK: omp.region.cont: ; preds = %[[VAL_78:.*]]
+// CHECK: br label %[[VAL_79:.*]]
+// CHECK: taskgroup.exit: ; preds = %[[VAL_80:.*]]
+// CHECK: call void @__kmpc_end_taskgroup(ptr @1, i32 %[[VAL_69]])
+// CHECK: ret void
+// CHECK: omp.taskloop.start: ; preds = %[[VAL_76]]
+// CHECK: br label %[[VAL_81:.*]]
+// CHECK: codeRepl: ; preds = %[[VAL_77]]
+// CHECK: %[[VAL_82:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_65]], i32 0, i32 0
+// CHECK: store i64 1, ptr %[[VAL_82]], align 4
+// CHECK: %[[VAL_83:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_65]], i32 0, i32 1
+// CHECK: store i64 100, ptr %[[VAL_83]], align 4
+// CHECK: %[[VAL_84:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_65]], i32 0, i32 2
+// CHECK: store i64 1, ptr %[[VAL_84]], align 4
+// CHECK: %[[VAL_85:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_65]], i32 0, i32 3
+// CHECK: store ptr %[[VAL_72]], ptr %[[VAL_85]], align 8
+// CHECK: %[[VAL_86:.*]] = call i32 @__kmpc_global_thread_num(ptr @1)
+// CHECK: %[[VAL_87:.*]] = call ptr @__kmpc_omp_task_alloc(ptr @1, i32 %[[VAL_86]], i32 1, i64 40, i64 32, ptr @_QPtest2..omp_par)
+// CHECK: %[[VAL_88:.*]] = load ptr, ptr %[[VAL_87]], align 8
+// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 1 %[[VAL_88]], ptr align 1 %[[VAL_65]], i64 32, i1 false)
+// CHECK: %[[VAL_89:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_88]], i32 0, i32 0
+// CHECK: %[[VAL_90:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_88]], i32 0, i32 1
+// CHECK: %[[VAL_91:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_88]], i32 0, i32 2
+// CHECK: %[[VAL_92:.*]] = load i64, ptr %[[VAL_91]], align 4
+// CHECK: call void @__kmpc_taskloop(ptr @1, i32 %[[VAL_86]], ptr %[[VAL_87]], i32 1, ptr %[[VAL_89]], ptr %[[VAL_90]], i64 %[[VAL_92]], i32 1, i32 0, i64 0, ptr @omp_taskloop_dup.1)
+// CHECK: br label %[[VAL_78]]
+// CHECK: taskloop.exit: ; preds = %[[VAL_81]]
+// CHECK: br label %[[VAL_80]]
+
+// CHECK-LABEL: define internal void @_QPtest2..omp_par(
+// CHECK: taskloop.alloca:
+// CHECK: %[[VAL_93:.*]] = load ptr, ptr %[[VAL_94:.*]], align 8
+// CHECK: %[[VAL_95:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_93]], i32 0, i32 0
+// CHECK: %[[VAL_96:.*]] = load i64, ptr %[[VAL_95]], align 4
+// CHECK: %[[VAL_97:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_93]], i32 0, i32 1
+// CHECK: %[[VAL_98:.*]] = load i64, ptr %[[VAL_97]], align 4
+// CHECK: %[[VAL_99:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_93]], i32 0, i32 2
+// CHECK: %[[VAL_100:.*]] = load i64, ptr %[[VAL_99]], align 4
+// CHECK: %[[VAL_101:.*]] = getelementptr { i64, i64, i64, ptr }, ptr %[[VAL_93]], i32 0, i32 3
+// CHECK: %[[VAL_102:.*]] = load ptr, ptr %[[VAL_101]], align 8, !align !1
+// CHECK: %[[VAL_103:.*]] = alloca i32, align 4
+// CHECK: br label %[[VAL_104:.*]]
+// CHECK: taskloop.body: ; preds = %[[VAL_105:.*]]
+// CHECK: %[[VAL_106:.*]] = getelementptr { i32 }, ptr %[[VAL_102]], i32 0, i32 0
+// CHECK: br label %[[VAL_107:.*]]
+// CHECK: omp.taskloop.region: ; preds = %[[VAL_104]]
+// CHECK: br label %[[VAL_108:.*]]
+// CHECK: omp_loop.preheader: ; preds = %[[VAL_107]]
+// CHECK: %[[VAL_109:.*]] = sub i64 %[[VAL_98]], %[[VAL_96]]
+// CHECK: %[[VAL_110:.*]] = sdiv i64 %[[VAL_109]], 1
+// CHECK: %[[VAL_111:.*]] = add i64 %[[VAL_110]], 1
+// CHECK: %[[VAL_112:.*]] = trunc i64 %[[VAL_111]] to i32
+// CHECK: %[[VAL_113:.*]] = trunc i64 %[[VAL_96]] to i32
+// CHECK: br label %[[VAL_114:.*]]
+// CHECK: omp_loop.header: ; preds = %[[VAL_115:.*]], %[[VAL_108]]
+// CHECK: %[[VAL_116:.*]] = phi i32 [ 0, %[[VAL_108]] ], [ %[[VAL_117:.*]], %[[VAL_115]] ]
+// CHECK: br label %[[VAL_118:.*]]
+// CHECK: omp_loop.cond: ; preds = %[[VAL_114]]
+// CHECK: %[[VAL_119:.*]] = icmp ult i32 %[[VAL_116]], %[[VAL_112]]
+// CHECK: br i1 %[[VAL_119]], label %[[VAL_120:.*]], label %[[VAL_121:.*]]
+// CHECK: omp_loop.exit: ; preds = %[[VAL_118]]
+// CHECK: br label %[[VAL_122:.*]]
+// CHECK: omp_loop.after: ; preds = %[[VAL_121]]
+// ...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/174815
More information about the llvm-branch-commits
mailing list