[MLIR] Change return type of ParallelOp::getInductionVars to ValueRange.

[pifon2a]

Summary

The current return type sometimes leads to code like to_vector<2>(ValueRange(loop.getInductionIvs())). I think it would be nice to shorten it. Users who need access to Block::BlockArgListType (if there are any), can always call getBody()->getArguments(); if needed.

Also remove getNumInductionVars(), since there is getNumLoops().

[mehdi_amini]

Can you motivate the change "Change return type of ParallelOp::getInductionVars to ValueRange" in the description ?

[pifon2a]

@mehdi_amini done, thanks!

[mehdi_amini]

Thanks!

[herhut]

Nice! Thanks.

mlir/include/mlir/Dialect/LoopOps/LoopOps.td

def ParallelOp : Loop_Op<"parallel",
   let builders = [
     OpBuilder<"Builder *builder, OperationState &result, "
               "ValueRange lowerBounds, ValueRange upperBounds, "
               "ValueRange steps, ValueRange initVals = {}">,
   ];
 
   let extraClassDeclaration = [{
     Block *getBody() { return &region().front(); }
-    unsigned getNumInductionVars() {
-      return getBody()->getNumArguments();
-    }
-    Block::BlockArgListType getInductionVars() {
+    ValueRange getInductionVars() {
       return getBody()->getArguments();
     }
     unsigned getNumLoops() { return step().size(); }
     unsigned getNumReductions() { return initVals().size(); }
   }];
 }
 
 def ReduceOp : Loop_Op<"reduce", [HasParent<"ParallelOp">]> {

mlir/lib/Dialect/GPU/Transforms/ParallelLoopMapper.cpp

 //===- ParallelLoopMapper.cpp - Utilities for mapping parallel loops to GPU =//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
static void mapParallelOp(ParallelOp parallelOp,
   // Do not try to add a mapping to already mapped loops or nested loops.
   if (parallelOp.getAttr(getMappingAttrName()) ||
       ((mappingLevel == MapGrid) && parallelOp.getParentOfType<ParallelOp>()))
     return;
 
   MLIRContext *ctx = parallelOp.getContext();
   Builder b(ctx);
   SmallVector<ParallelLoopDimMapping, 4> attrs;
-  attrs.reserve(parallelOp.getNumInductionVars());
-  for (int i = 0, e = parallelOp.getNumInductionVars(); i < e; ++i) {
+  attrs.reserve(parallelOp.getNumLoops());
+  for (int i = 0, e = parallelOp.getNumLoops(); i < e; ++i) {
     attrs.push_back(getParallelLoopDimMappingAttr(
         getHardwareIdForMapping(mappingLevel, i), b.getDimIdentityMap(),
         b.getDimIdentityMap()));
   }
   setMappingAttr(parallelOp, attrs);
   ++mappingLevel;
   // Parallel loop operations are immediately nested, so do not use
   // walk but just iterate over the operations.