[Outlining] Changing stringify values to 32-bit (#5832)

The LLVM suffix tree expects to be provided with a vector of 32-bit unsigned integers. This PR makes it easier to integrate our wasm program string with the suffix tree.

Because the range of possible values is reduced from 2^64 to 2^32, a signed integer was added to manage the next separator value and ensure we're using every possible negative number.

2 files changed, 10 insertions, 6 deletions

diff --git a/src/passes/hash-stringify-walker.cpp b/src/passes/hash-stringify-walker.cpp
index 8d76aeec9..3df01b290 100644
--- a/src/passes/hash-stringify-walker.cpp
+++ b/src/passes/hash-stringify-walker.cpp
@@ -55,8 +55,9 @@ bool StringifyEquator::operator()(Expression* lhs, Expression* rhs) const {
 void HashStringifyWalker::addUniqueSymbol() {
   // Use a negative value to distinguish symbols for separators from symbols
   // for Expressions
-  hashString.push_back((uint64_t)-nextVal);
-  nextVal++;
+  assert((uint32_t)nextSeparatorVal >= nextVal);
+  hashString.push_back((uint32_t)nextSeparatorVal);
+  nextSeparatorVal--;
 }
 
 void HashStringifyWalker::visitExpression(Expression* curr) {
diff --git a/src/passes/stringify-walker.h b/src/passes/stringify-walker.h
index 9563000d0..7e18d3d9e 100644
--- a/src/passes/stringify-walker.h
+++ b/src/passes/stringify-walker.h
@@ -113,17 +113,20 @@ struct StringifyEquator {
 
 struct HashStringifyWalker : public StringifyWalker<HashStringifyWalker> {
   // After calling walkModule, this vector contains the result of encoding a
-  // wasm module as a string of uint64_t values. Each value represents either an
+  // wasm module as a string of uint32_t values. Each value represents either an
   // Expression or a separator to mark the end of control flow.
-  std::vector<uint64_t> hashString;
+  std::vector<uint32_t> hashString;
   // A monotonic counter used to ensure that unique expressions in the
   // module are assigned a unique value in the hashString
-  uint64_t nextVal = 0;
+  uint32_t nextVal = 0;
+  // A monotonic counter used to ensure that each separator in the
+  // module is assigned a unique value in the hashString
+  int32_t nextSeparatorVal = -1;
   // Contains a mapping of expression pointer to value to ensure we
   // use the same value for matching expressions. A custom hasher and
   // equator is provided in order to separate out evaluation of the if-condition
   // when evaluating if expressions.
-  std::unordered_map<Expression*, uint64_t, StringifyHasher, StringifyEquator>
+  std::unordered_map<Expression*, uint32_t, StringifyHasher, StringifyEquator>
     exprToCounter;
 
   void addUniqueSymbol();