[MLIR][OpenMP] Add omp.loop_nest operation

mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td

@@ -524,6 +524,70 @@ def SingleOp : OpenMP_Op<"single", [AttrSizedOperandSegments]> {
   let hasVerifier = 1;
 }
 
+//===----------------------------------------------------------------------===//
+// Loop Nest
+//===----------------------------------------------------------------------===//
+
+def LoopNestOp : OpenMP_Op<"loop_nest", [SameVariadicOperandSize,
+                        AllTypesMatch<["lowerBound", "upperBound", "step"]>,
+                        ParentOneOf<["DistributeOp", "SimdLoopOp", "TaskloopOp",
+                                     "WsloopOp"]>,
+                        RecursiveMemoryEffects]> {
+  let summary = "rectangular loop nest";
+  let description = [{
+    This operation represents a collapsed rectangular loop nest. For each
+    rectangular loop of the nest represented by an instance of this operation,
+    lower and upper bounds, as well as a step variable, must be defined.
+
+    The lower and upper bounds specify a half-open range: the range includes the
+    lower bound but does not include the upper bound. If the `inclusive`
+    attribute is specified then the upper bound is also included.
+
+    The body region can contain any number of blocks. The region is terminated
+    by an `omp.yield` instruction without operands. The induction variables,
+    represented as entry block arguments to the loop nest operation's single
+    region, match the types of the `lowerBound`, `upperBound` and `step`
+    arguments.
+
+    ```mlir
+    omp.loop_nest (%i1, %i2) : i32 = (%c0, %c0) to (%c10, %c10) step (%c1, %c1) {
+      %a = load %arrA[%i1, %i2] : memref<?x?xf32>
+      %b = load %arrB[%i1, %i2] : memref<?x?xf32>
+      %sum = arith.addf %a, %b : f32
+      store %sum, %arrC[%i1, %i2] : memref<?x?xf32>
+      omp.yield
+    }
+    ```
+
+    This is a temporary simplified definition of a loop based on existing OpenMP
+    loop operations intended to serve as a stopgap solution until the long-term
+    representation of canonical loops is defined. Specifically, this operation
+    is intended to serve as a unique source for loop information during the
+    transition to making `omp.distribute`, `omp.simdloop`, `omp.taskloop` and
+    `omp.wsloop` wrapper operations. It is not intended to help with the
+    addition of support for loop transformations, non-rectangular loops and
+    non-perfectly nested loops.
+  }];
+
+  let arguments = (ins Variadic<IntLikeType>:$lowerBound,
+                       Variadic<IntLikeType>:$upperBound,
+                       Variadic<IntLikeType>:$step,
+                       UnitAttr:$inclusive);
+
+  let regions = (region AnyRegion:$region);
+
+  let extraClassDeclaration = [{
+    /// Returns the number of loops in the loop nest.
+    unsigned getNumLoops() { return getLowerBound().size(); }
+
+    /// Returns the induction variables of the loop nest.
+    ArrayRef<BlockArgument> getIVs() { return getRegion().getArguments(); }
+  }];
+
+  let hasCustomAssemblyFormat = 1;
+  let hasVerifier = 1;
+}
+
 //===----------------------------------------------------------------------===//
 // 2.9.2 Workshare Loop Construct
 //===----------------------------------------------------------------------===//
@@ -743,7 +807,7 @@ def SimdLoopOp : OpenMP_Op<"simdloop", [AttrSizedOperandSegments,
 
 def YieldOp : OpenMP_Op<"yield",
     [Pure, ReturnLike, Terminator,
-     ParentOneOf<["WsloopOp", "DeclareReductionOp",
+     ParentOneOf<["LoopNestOp", "WsloopOp", "DeclareReductionOp",
      "AtomicUpdateOp", "SimdLoopOp", "PrivateClauseOp"]>]> {
   let summary = "loop yield and termination operation";
   let description = [{

mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp

@@ -1862,6 +1862,77 @@ LogicalResult TaskloopOp::verify() {
   return success();
 }
 
+//===----------------------------------------------------------------------===//
+// LoopNestOp
+//===----------------------------------------------------------------------===//
+
+ParseResult LoopNestOp::parse(OpAsmParser &parser, OperationState &result) {
+  // Parse an opening `(` followed by induction variables followed by `)`
+  SmallVector<OpAsmParser::Argument> ivs;
+  SmallVector<OpAsmParser::UnresolvedOperand> lbs, ubs;
+  Type loopVarType;
+  if (parser.parseArgumentList(ivs, OpAsmParser::Delimiter::Paren) ||
+      parser.parseColonType(loopVarType) ||
+      // Parse loop bounds.
+      parser.parseEqual() ||
+      parser.parseOperandList(lbs, ivs.size(), OpAsmParser::Delimiter::Paren) ||
+      parser.parseKeyword("to") ||
+      parser.parseOperandList(ubs, ivs.size(), OpAsmParser::Delimiter::Paren))
+    return failure();
+
+  for (auto &iv : ivs)
+    iv.type = loopVarType;
+
+  // Parse "inclusive" flag.
+  if (succeeded(parser.parseOptionalKeyword("inclusive")))
+    result.addAttribute("inclusive",
+                        UnitAttr::get(parser.getBuilder().getContext()));
+
+  // Parse step values.
+  SmallVector<OpAsmParser::UnresolvedOperand> steps;
+  if (parser.parseKeyword("step") ||
+      parser.parseOperandList(steps, ivs.size(), OpAsmParser::Delimiter::Paren))
+    return failure();
+
+  // Parse the body.
+  Region *region = result.addRegion();
+  if (parser.parseRegion(*region, ivs))
+    return failure();
+
+  // Resolve operands.
+  if (parser.resolveOperands(lbs, loopVarType, result.operands) ||
+      parser.resolveOperands(ubs, loopVarType, result.operands) ||
+      parser.resolveOperands(steps, loopVarType, result.operands))
+    return failure();
+
+  // Parse the optional attribute list.
+  return parser.parseOptionalAttrDict(result.attributes);
+}
+
+void LoopNestOp::print(OpAsmPrinter &p) {
+  Region &region = getRegion();
+  auto args = region.getArguments();
+  p << " (" << args << ") : " << args[0].getType() << " = (" << getLowerBound()
+    << ") to (" << getUpperBound() << ") ";
+  if (getInclusive())
+    p << "inclusive ";
+  p << "step (" << getStep() << ") ";
+  p.printRegion(region, /*printEntryBlockArgs=*/false);
+}
+
+LogicalResult LoopNestOp::verify() {
+  if (getLowerBound().size() != getIVs().size())
+    return emitOpError() << "number of range arguments and IVs do not match";
+
+  for (auto [lb, iv] : llvm::zip_equal(getLowerBound(), getIVs())) {
+    if (lb.getType() != iv.getType())
+      return emitOpError()
+             << "range argument type does not match corresponding IV type";
+  }
+
+  return success();
+}
+
 //===----------------------------------------------------------------------===//
 // WsloopOp
 //===----------------------------------------------------------------------===//

mlir/test/Dialect/OpenMP/invalid.mlir

@@ -87,6 +87,43 @@ func.func @proc_bind_once() {
 
 // -----
 
+func.func @invalid_parent(%lb : index, %ub : index, %step : index) {
+  // expected-error@+1 {{op expects parent op to be one of 'omp.distribute, omp.simdloop, omp.taskloop, omp.wsloop'}}
+  omp.loop_nest (%iv) : index = (%lb) to (%ub) step (%step) {
+    omp.yield
+  }
+}
+
+// -----
+
+func.func @type_mismatch(%lb : index, %ub : index, %step : index) {
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // expected-error@+1 {{range argument type does not match corresponding IV type}}
+    "omp.loop_nest" (%lb, %ub, %step) ({
+    ^bb0(%iv2: i32):
+      omp.yield
+    }) : (index, index, index) -> ()
+    omp.yield
+  }
+}
+
+// -----
+
+func.func @iv_number_mismatch(%lb : index, %ub : index, %step : index) {
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // expected-error@+1 {{number of range arguments and IVs do not match}}
+    "omp.loop_nest" (%lb, %ub, %step) ({
+    ^bb0(%iv1 : index, %iv2 : index):
+      omp.yield
+    }) : (index, index, index) -> ()
+    omp.yield
+  }
+}
+
+// -----
+
 func.func @inclusive_not_a_clause(%lb : index, %ub : index, %step : index) {
   // expected-error @below {{expected 'for'}}
   omp.wsloop nowait inclusive

mlir/test/Dialect/OpenMP/ops.mlir

@@ -133,6 +133,85 @@ func.func @omp_parallel_pretty(%data_var : memref<i32>, %if_cond : i1, %num_thre
   return
 }
 
+// CHECK-LABEL: omp_loop_nest
+func.func @omp_loop_nest(%lb : index, %ub : index, %step : index) -> () {
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}) to (%{{.*}}) step (%{{.*}})
+    "omp.loop_nest" (%lb, %ub, %step) ({
+    ^bb0(%iv2: index):
+      omp.yield
+    }) : (index, index, index) -> ()
+    omp.yield
+  }
+
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}) to (%{{.*}}) inclusive step (%{{.*}})
+    "omp.loop_nest" (%lb, %ub, %step) ({
+    ^bb0(%iv2: index):
+      omp.yield
+    }) {inclusive} : (index, index, index) -> ()
+    omp.yield
+  }
+
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}, %{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}, %{{.*}}) to (%{{.*}}, %{{.*}}) step (%{{.*}}, %{{.*}})
+    "omp.loop_nest" (%lb, %lb, %ub, %ub, %step, %step) ({
+    ^bb0(%iv2: index, %iv3: index):
+      omp.yield
+    }) : (index, index, index, index, index, index) -> ()
+    omp.yield
+  }
+
+  return
+}
+
+// CHECK-LABEL: omp_loop_nest_pretty
+func.func @omp_loop_nest_pretty(%lb : index, %ub : index, %step : index) -> () {
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}) to (%{{.*}}) step (%{{.*}})
+    omp.loop_nest (%iv2) : index = (%lb) to (%ub) step (%step) {
+      omp.yield
+    }
+    omp.yield
+  }
+
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}) to (%{{.*}}) inclusive step (%{{.*}})
+    omp.loop_nest (%iv2) : index = (%lb) to (%ub) inclusive step (%step) {
+      omp.yield
+    }
+    omp.yield
+  }
+
+  // TODO Remove induction variables from omp.wsloop.
+  omp.wsloop for (%iv) : index = (%lb) to (%ub) step (%step) {
+    // CHECK: omp.loop_nest
+    // CHECK-SAME: (%{{.*}}) : index =
+    // CHECK-SAME: (%{{.*}}, %{{.*}}) to (%{{.*}}, %{{.*}}) step (%{{.*}}, %{{.*}})
+    omp.loop_nest (%iv2, %iv3) : index = (%lb, %lb) to (%ub, %ub) step (%step, %step) {
+      omp.yield
+    }
+    omp.yield
+  }
+
+  return
+}
+
 // CHECK-LABEL: omp_wsloop
 func.func @omp_wsloop(%lb : index, %ub : index, %step : index, %data_var : memref<i32>, %linear_var : i32, %chunk_var : i32) -> () {