[Mlir-commits] [mlir] [mlir][SCF] Modernize `coalesceLoops` method to handle `scf.for` loops with iter_args (PR #87019)

[Oleksandr Alex Zinenko]

================
@@ -2765,3 +2765,49 @@ mlir::affine::separateFullTiles(MutableArrayRef<AffineForOp> inputNest,
 
   return success();
 }
+
+LogicalResult affine::coalescePerfectlyNestedAffineLoops(AffineForOp op) {
+  LogicalResult result(failure());
+  SmallVector<AffineForOp> loops;
+  getPerfectlyNestedLoops(loops, op);
+
+  // Look for a band of loops that can be coalesced, i.e. perfectly nested
+  // loops with bounds defined above some loop.
+  // 1. For each loop, find above which parent loop its operands are
+  // defined.
+  SmallVector<unsigned, 4> operandsDefinedAbove(loops.size());
----------------
ftynse wrote:

Nit: drop the explicit number of stack elements.

https://github.com/llvm/llvm-project/pull/87019