Fix Murmur3 hash

Author: pmenon

src/codegen/hash.cpp

@@ -169,8 +169,8 @@ llvm::Value *Hash::ComputeCRC32Hash(CodeGen &codegen,
   // Now hash the strings
   for (auto &varlen : varlens) {
     llvm::Value *len = codegen->CreateZExt(varlen.len, codegen.Int64Type());
-    crc = codegen.Call(RuntimeFunctionsProxy::HashCrc64,
-                       {varlen.val, len, crc});
+    crc =
+        codegen.Call(RuntimeFunctionsProxy::HashCrc64, {varlen.val, len, crc});
   }
 
   ///
@@ -181,7 +181,7 @@ llvm::Value *Hash::ComputeCRC32Hash(CodeGen &codegen,
 // single value
 llvm::Value *Hash::ComputeMurmur3Hash(
     CodeGen &codegen, const std::vector<llvm::Value *> &numerics,
-    const std::vector<Hash::Varlen> &varlen_buffers) {
+    const std::vector<Hash::Varlen> &varlens) {
   // The magic constants used in Murmur3's final 64-bit avalanche mix
   static constexpr uint64_t kMurmur3C1 = 0xff51afd7ed558ccdLLU;
   static constexpr uint64_t kMurmur3C2 = 0xc4ceb9fe1a85ec53LLU;
@@ -223,8 +223,16 @@ llvm::Value *Hash::ComputeMurmur3Hash(
     }
   }
 
-  if (!varlen_buffers.empty()) {
-    throw Exception{"Cannot perform a vectorized Murmur3 hash on strings"};
+  if (hash == nullptr) {
+    // No numerics in the hash
+    hash = codegen.Const64(0);
+  }
+
+  // Now hash the strings
+  for (auto &varlen : varlens) {
+    llvm::Value *len = codegen->CreateZExt(varlen.len, codegen.Int64Type());
+    std::vector<llvm::Value *> args = {varlen.val, len, hash};
+    hash = codegen.Call(RuntimeFunctionsProxy::HashMurmur3, args);
   }
 
   return hash;

src/codegen/operator/table_scan_translator.cpp

@@ -221,7 +221,7 @@ void TableScanTranslator::ScanConsumer::SetupRowBatch(
   for (oid_t col_idx = 0; col_idx < output_col_ids.size(); col_idx++) {
     auto *attribute = ais[output_col_ids[col_idx]];
     LOG_TRACE("Adding attribute '%s.%s' (%p) into row batch",
-              scan_plan.GetTable()->GetName().c_str(), attribute->name.c_str(),
+              plan_.GetTable()->GetName().c_str(), attribute->name.c_str(),
               attribute);
     batch.AddAttribute(attribute, &access[col_idx]);
   }

src/codegen/proxy/runtime_functions_proxy.cpp

@@ -26,6 +26,7 @@ DEFINE_TYPE(ColumnLayoutInfo, "peloton::ColumnLayoutInfo", col_start_ptr,
 DEFINE_TYPE(AbstractExpression, "peloton::expression::AbstractExpression",
             opaque);
 
+DEFINE_METHOD(peloton::codegen, RuntimeFunctions, HashMurmur3);
 DEFINE_METHOD(peloton::codegen, RuntimeFunctions, HashCrc64);
 DEFINE_METHOD(peloton::codegen, RuntimeFunctions, GetTileGroup);
 DEFINE_METHOD(peloton::codegen, RuntimeFunctions, GetTileGroupLayout);

src/include/codegen/hash.h

@@ -54,7 +54,7 @@ class Hash {
 
   static llvm::Value *ComputeMurmur3Hash(
       CodeGen &codegen, const std::vector<llvm::Value *> &numerics,
-      const std::vector<Hash::Varlen> &varlen_buffers);
+      const std::vector<Hash::Varlen> &varlens);
 };
 
 }  // namespace codegen

src/include/codegen/proxy/runtime_functions_proxy.h

@@ -34,6 +34,7 @@ PROXY(AbstractExpression) {
 };
 
 PROXY(RuntimeFunctions) {
+  DECLARE_METHOD(HashMurmur3);
   DECLARE_METHOD(HashCrc64);
   DECLARE_METHOD(GetTileGroup);
   DECLARE_METHOD(GetTileGroupLayout);