Raghav/inline structs refactor2

changelogs/unreleased/raghav__inline-structs-refactor2.yaml

@@ -0,0 +1,2 @@
+changed:
+  - Refactored LLZKInlineStructs pass to allow inlining on demand

include/llzk/Transforms/LLZKInlineStructsPass.h

@@ -0,0 +1,26 @@
+//===-- LLZKInlineStructsPass.h ---------------------------------*- C++ -*-===//
+//
+// Part of the LLZK Project, under the Apache License v2.0.
+// See LICENSE.txt for license information.
+// Copyright 2025 Veridise Inc.
+// SPDX-License-Identifier: Apache-2.0
+//
+//===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include <llzk/Dialect/Struct/IR/Ops.h>
+
+#include <mlir/IR/SymbolTable.h>
+#include <mlir/Support/LogicalResult.h>
+
+/// Maps caller struct to callees that should be inlined. The outer SmallVector preserves the
+/// ordering from the bottom-up traversal that builds the InliningPlan so performing inlining
+/// in the order given will not lose any or require doing any more than once.
+/// Note: Applying in the opposite direction would reduce making repeated clones of the ops within
+/// the inlined struct functions (as they are inlined further and further up the tree) but that
+/// would require updating some mapping in the plan along the way to ensure it's done properly.
+using InliningPlan = mlir::SmallVector<
+    std::pair<llzk::component::StructDefOp, mlir::SmallVector<llzk::component::StructDefOp>>>;
+
+mlir::LogicalResult performInlining(mlir::SymbolTableCollection &tables, InliningPlan &plan);