[analysis][NFC] Create a TransferFunction concept (WebAssembly#6033)

Factor the static assertions for transfer functions out into a new
transfer-function.h header. The concept requires the `getDependents` method to
return an input range of basic blocks, and to satisfy that requirement, fix up
_indirect_ptr_iterator in cfg-impl.h so that it is a proper iterator. Remove
part of the lattice fuzzer that was using a placeholder transfer function in a
way that does not satisfy the new type constraints; most of that code will be
overhauled in the future anyway.

Author: tlively

src/analysis/cfg-impl.h

@@ -27,7 +27,7 @@ namespace wasm::analysis {
 template<typename T> struct _indirect_ptr_iterator {
   using iterator_category = std::random_access_iterator_tag;
   using value_type = T;
-  using different_type = off_t;
+  using difference_type = off_t;
   using reference = const T&;
   using pointer = const T*;
 
@@ -75,6 +75,10 @@ template<typename T> struct _indirect_ptr_iterator {
     return it;
   }
 
+  off_t operator-(const _indirect_ptr_iterator& other) const {
+    return ptr - other.ptr;
+  }
+
   bool operator==(const _indirect_ptr_iterator& other) const {
     return ptr == other.ptr;
   }
@@ -98,8 +102,17 @@ template<typename T> struct _indirect_ptr_iterator {
   bool operator>=(const _indirect_ptr_iterator& other) const {
     return ptr >= other.ptr;
   }
+
+  friend _indirect_ptr_iterator operator+(off_t n,
+                                          const _indirect_ptr_iterator& it) {
+    return it + n;
+  }
 };
 
+#if __cplusplus >= 202002L
+static_assert(std::random_access_iterator<_indirect_ptr_iterator<int>>);
+#endif
+
 template<typename T>
 _indirect_ptr_iterator<T> operator+(int n,
                                     const _indirect_ptr_iterator<T>& it) {

src/analysis/lattice.h

@@ -17,6 +17,10 @@
 #ifndef wasm_analysis_lattice_h
 #define wasm_analysis_lattice_h
 
+#if __cplusplus >= 202002L
+#include <concepts>
+#endif
+
 namespace wasm::analysis {
 
 enum LatticeComparison { NO_RELATION, EQUAL, LESS, GREATER };
@@ -35,8 +39,6 @@ inline LatticeComparison reverseComparison(LatticeComparison comparison) {
 
 #if __cplusplus >= 202002L
 
-#include <concepts>
-
 template<typename L>
 concept Lattice = requires(const L& lattice,
                            const typename L::Element& constElem,

src/analysis/monotone-analyzer-impl.h

@@ -30,29 +30,30 @@ template<Lattice L> inline void BlockState<L>::print(std::ostream& os) {
   os << std::endl;
 }
 
-template<Lattice L, typename TransferFunction>
-inline MonotoneCFGAnalyzer<L, TransferFunction>::MonotoneCFGAnalyzer(
-  L& lattice, TransferFunction& transferFunction, CFG& cfg)
-  : lattice(lattice), transferFunction(transferFunction), cfg(cfg) {
+template<Lattice L, TransferFunction TxFn>
+inline MonotoneCFGAnalyzer<L, TxFn>::MonotoneCFGAnalyzer(L& lattice,
+                                                         TxFn& txfn,
+                                                         CFG& cfg)
+  : lattice(lattice), txfn(txfn), cfg(cfg) {
 
   // Construct BlockStates for each BasicBlock.
   for (auto it = cfg.begin(); it != cfg.end(); it++) {
     stateBlocks.emplace_back(&(*it), lattice);
   }
 }
 
-template<Lattice L, typename TransferFunction>
-inline void MonotoneCFGAnalyzer<L, TransferFunction>::evaluateFunctionEntry(
-  Function* func) {
-  transferFunction.evaluateFunctionEntry(func, stateBlocks[0].inputState);
+template<Lattice L, TransferFunction TxFn>
+inline void
+MonotoneCFGAnalyzer<L, TxFn>::evaluateFunctionEntry(Function* func) {
+  txfn.evaluateFunctionEntry(func, stateBlocks[0].inputState);
 }
 
-template<Lattice L, typename TransferFunction>
-inline void MonotoneCFGAnalyzer<L, TransferFunction>::evaluate() {
+template<Lattice L, TransferFunction TxFn>
+inline void MonotoneCFGAnalyzer<L, TxFn>::evaluate() {
   std::queue<const BasicBlock*> worklist;
 
   // Transfer function enqueues the work in some order which is efficient.
-  transferFunction.enqueueWorklist(cfg, worklist);
+  txfn.enqueueWorklist(cfg, worklist);
 
   while (!worklist.empty()) {
     BlockState<L>& currBlockState = stateBlocks[worklist.front()->getIndex()];
@@ -63,10 +64,10 @@ inline void MonotoneCFGAnalyzer<L, TransferFunction>::evaluate() {
     // to arrive at the expression's state. The beginning and end states of the
     // CFG block will be updated.
     typename L::Element outputState = currBlockState.inputState;
-    transferFunction.transfer(currBlockState.cfgBlock, outputState);
+    txfn.transfer(currBlockState.cfgBlock, outputState);
 
     // Propagate state to dependents of currBlockState.
-    for (auto& dep : transferFunction.getDependents(currBlockState.cfgBlock)) {
+    for (auto& dep : txfn.getDependents(currBlockState.cfgBlock)) {
       // If we need to change the input state of a dependent, we need
       // to enqueue the dependent to recalculate it.
       if (stateBlocks[dep.getIndex()].inputState.makeLeastUpperBound(
@@ -77,28 +78,28 @@ inline void MonotoneCFGAnalyzer<L, TransferFunction>::evaluate() {
   }
 }
 
-template<Lattice L, typename TransferFunction>
-inline void MonotoneCFGAnalyzer<L, TransferFunction>::collectResults() {
+template<Lattice L, TransferFunction TxFn>
+inline void MonotoneCFGAnalyzer<L, TxFn>::collectResults() {
   for (BlockState currBlockState : stateBlocks) {
     typename L::Element inputStateCopy = currBlockState.inputState;
 
     // The transfer function generates the final set of states and uses it to
     // produce useful information. For example, in reaching definitions
     // analysis, these final states are used to populate a mapping of
     // local.get's to a set of local.set's that affect its value.
-    transferFunction.collectResults(currBlockState.cfgBlock, inputStateCopy);
+    txfn.collectResults(currBlockState.cfgBlock, inputStateCopy);
   }
 }
 
 // Currently prints both the basic information and intermediate states of each
 // BlockState.
-template<Lattice L, typename TransferFunction>
-inline void MonotoneCFGAnalyzer<L, TransferFunction>::print(std::ostream& os) {
+template<Lattice L, TransferFunction TxFn>
+inline void MonotoneCFGAnalyzer<L, TxFn>::print(std::ostream& os) {
   os << "CFG Analyzer" << std::endl;
   for (auto state : stateBlocks) {
     state.print(os);
     typename L::Element temp = state.inputState;
-    transferFunction.print(os, state.cfgBlock, temp);
+    txfn.print(os, state.cfgBlock, temp);
   }
   os << "End" << std::endl;
 }

src/analysis/monotone-analyzer.h

@@ -7,6 +7,7 @@
 
 #include "cfg.h"
 #include "lattice.h"
+#include "transfer-function.h"
 
 namespace wasm::analysis {
 
@@ -26,76 +27,16 @@ template<Lattice L> struct BlockState {
   void print(std::ostream& os);
 };
 
-// A transfer function using a lattice <Lattice> is required to have the
-// following methods:
-
-// Lattice::Element transfer(const BasicBlock* cfgBlock, Lattice::Element&
-// inputState);
-
-// This function takes in a pointer to a CFG BasicBlock and the input state
-// associated with it and modifies the input state in-place into the ouptut
-// state for the basic block by applying the analysis transfer function to each
-// expression in the CFG BasicBlock. Starting with the input state, the transfer
-// function is used to change the state to new intermediate states based on each
-// expression until we reach the output state. The outuput state will be
-// propagated to dependents of the CFG BasicBlock by the worklist algorithm in
-// MonotoneCFGAnalyzer.
-
-template<typename TransferFunction, Lattice L>
-constexpr bool has_transfer =
-  std::is_invocable_r<void,
-                      decltype(&TransferFunction::transfer),
-                      TransferFunction,
-                      const BasicBlock*,
-                      typename L::Element&>::value;
-
-// void enqueueWorklist(CFG&, std::queue<const BasicBlock*>& value);
-
-// Loads CFG BasicBlocks in some order into the worklist. Custom specifying the
-// order for each analysis brings performance savings. For example, when doing a
-// backward analysis, loading the BasicBlocks in reverse order will lead to less
-// state propagations, and therefore better performance. The opposite is true
-// for a forward analysis.
-
-template<typename TransferFunction, Lattice L>
-constexpr bool has_enqueueWorklist =
-  std::is_invocable_r<void,
-                      decltype(&TransferFunction::enqueueWorklist),
-                      TransferFunction,
-                      CFG&,
-                      std::queue<const BasicBlock*>&>::value;
-
-// BasicBlock::BasicBlockIterable getDependents(const BasicBlock* currBlock);
-
-// Returns an iterable to the CFG BasicBlocks which depend on currBlock for
-// information (e.g. predecessors in a backward analysis). Used to select which
-// blocks to propagate to after applying the transfer function to a block.
-
-template<typename TransferFunction>
-constexpr bool has_getDependents =
-  std::is_invocable_r<BasicBlock::BasicBlockIterable,
-                      decltype(&TransferFunction::getDependents),
-                      TransferFunction,
-                      const BasicBlock*>::value;
-
-// Combined TransferFunction assertions.
-template<typename TransferFunction, Lattice L>
-constexpr bool is_TransferFunction = has_transfer<TransferFunction, L> &&
-                                     has_enqueueWorklist<TransferFunction, L> &&
-                                     has_getDependents<TransferFunction>;
-
-template<Lattice L, typename TransferFunction> class MonotoneCFGAnalyzer {
-  static_assert(is_TransferFunction<TransferFunction, L>);
-
+template<Lattice L, TransferFunction TxFn> class MonotoneCFGAnalyzer {
   L& lattice;
-  TransferFunction& transferFunction;
+  TxFn& txfn;
   CFG& cfg;
   std::vector<BlockState<L>> stateBlocks;
 
 public:
   // Will constuct BlockState objects corresponding to BasicBlocks from the
   // given CFG.
-  MonotoneCFGAnalyzer(L& lattice, TransferFunction& transferFunction, CFG& cfg);
+  MonotoneCFGAnalyzer(L& lattice, TxFn& txfn, CFG& cfg);
 
   // Runs the worklist algorithm to compute the states for the BlockState graph.
   void evaluate();

src/analysis/transfer-function.h

@@ -0,0 +1,67 @@
+/*
+ * Copyright 2023 WebAssembly Community Group participants
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef wasm_analysis_transfer_function_h
+#define wasm_analysis_transfer_function_h
+
+#if __cplusplus >= 202002L
+#include <concepts>
+#include <iterator>
+#include <ranges>
+#endif
+
+#include <queue>
+
+#include "cfg.h"
+#include "lattice.h"
+
+namespace wasm::analysis {
+
+#if __cplusplus >= 202002L
+
+template<typename T>
+concept BasicBlockInputRange =
+  std::ranges::input_range<T> &&
+  std::is_same<std::ranges::range_value_t<T>, BasicBlock>::value;
+
+template<typename TxFn, typename L>
+concept TransferFunctionImpl = requires(TxFn& txfn,
+                                        const CFG& cfg,
+                                        BasicBlock* bb,
+                                        typename L::Element& elem,
+                                        std::queue<const BasicBlock*>& bbq) {
+  // Apply the transfer function to update a lattice element with information
+  // from a basic block.
+  { txfn.transfer(bb, elem) } noexcept -> std::same_as<void>;
+  // Initializes the worklist of basic blocks, which can affect performance
+  // depending on the direction of the analysis. TODO: Unlock performance
+  // benefits while exposing fewer implementation details.
+  { txfn.enqueueWorklist(cfg, bbq) } noexcept -> std::same_as<void>;
+  // Get a range over the basic blocks that depend on the given block.
+  { txfn.getDependents(bb) } noexcept -> BasicBlockInputRange;
+};
+
+#define TransferFunction TransferFunctionImpl<L>
+
+#else
+
+#define TransferFunction typename
+
+#endif // __cplusplus >= 202002L
+
+} // namespace wasm::analysis
+
+#endif // wasm_analysis_transfer_function_h

src/analysis/visitor-transfer-function.h

@@ -35,7 +35,8 @@ struct VisitorTransferFunc : public Visitor<SubType> {
 public:
   // Returns an iterable to all the BasicBlocks which depend on currBlock for
   // information.
-  BasicBlock::BasicBlockIterable getDependents(const BasicBlock* currBlock) {
+  BasicBlock::BasicBlockIterable
+  getDependents(const BasicBlock* currBlock) noexcept {
     if constexpr (Direction == AnalysisDirection::Backward) {
       return currBlock->preds();
     } else {
@@ -46,7 +47,8 @@ struct VisitorTransferFunc : public Visitor<SubType> {
   // Executes the transfer function on all the expressions of the corresponding
   // CFG node, starting with the node's input state, and changes the input state
   // to the final output state of the node in place.
-  void transfer(const BasicBlock* cfgBlock, typename L::Element& inputState) {
+  void transfer(const BasicBlock* cfgBlock,
+                typename L::Element& inputState) noexcept {
     // If the block is empty, we propagate the state by inputState =
     // outputState.
 
@@ -71,7 +73,8 @@ struct VisitorTransferFunc : public Visitor<SubType> {
   // analysis, we push all the blocks in order, while for backward analysis, we
   // push them in reverse order, so that later blocks are evaluated before
   // earlier ones.
-  void enqueueWorklist(CFG& cfg, std::queue<const BasicBlock*>& worklist) {
+  void enqueueWorklist(const CFG& cfg,
+                       std::queue<const BasicBlock*>& worklist) noexcept {
     if constexpr (Direction == AnalysisDirection::Backward) {
       for (auto it = cfg.rbegin(); it != cfg.rend(); ++it) {
         worklist.push(&(*it));