Method CodegenHashRow

be/src/exec/hash-table.cc:1026–1175

Codegen for hashing the current row. In the case with both string and non-string data (group by int_col, string_col), the IR looks like: define i32 @HashRow(%"class.impala::HashTableCtx"* %this_ptr, i8* %expr_values, i8* %expr_values_null) #34 { entry: %seed = call i32 @_ZNK6impala12HashTableCtx11GetHashSeedEv( %"class.impala::HashTableCtx"* %this_ptr) %hash = call i32 @CrcHash8(i8* %expr_values

//   ret i32 %hash_phi
// }
Status HashTableCtx::CodegenHashRow(LlvmCodeGen* codegen, bool use_murmur,
      const HashTableConfig& config, llvm::Function** fn) {
  const std::vector<ScalarExpr*>& exprs = config.build_exprs;
  const ScalarExprsResultsRowLayout& result_row_layout =
      config.build_exprs_results_row_layout;
  for (int i = 0; i < exprs.size(); ++i) {
    // Disable codegen for CHAR
    if (exprs[i]->type().type == TYPE_CHAR) {
      return Status("HashTableCtx::CodegenHashRow(): CHAR NYI");
    }
  }

  // Get types to generate function prototype
  llvm::PointerType* this_ptr_type = codegen->GetStructPtrType<HashTableCtx>();

  LlvmCodeGen::FnPrototype prototype(
      codegen, (use_murmur ? "MurmurHashRow" : "HashRow"), codegen->i32_type());
  prototype.AddArgument(LlvmCodeGen::NamedVariable("this_ptr", this_ptr_type));
  prototype.AddArgument(LlvmCodeGen::NamedVariable("expr_values", codegen->ptr_type()));
  prototype.AddArgument(
      LlvmCodeGen::NamedVariable("expr_values_null", codegen->ptr_type()));

  llvm::LLVMContext& context = codegen->context();
  LlvmBuilder builder(context);
  llvm::Value* args[3];
  *fn = prototype.GeneratePrototype(&builder, args);
  llvm::Value* this_arg = args[0];
  llvm::Value* expr_values = args[1];
  llvm::Value* expr_values_null = args[2];

  // Call GetHashSeed() to get seeds_[level_]
  llvm::Value* seed = codegen->CodegenCallFunction(
      &builder, IRFunction::HASH_TABLE_GET_HASH_SEED, this_arg, "seed");

  llvm::Value* hash_result = seed;
  const int var_result_offset = result_row_layout.var_results_begin_offset;
  const int expr_values_bytes_per_row = result_row_layout.expr_values_bytes_per_row;
  if (var_result_offset == -1) {
    // No variable length slots, just hash what is in 'expr_expr_values_cache_'
    if (expr_values_bytes_per_row > 0) {
      llvm::Function* hash_fn = use_murmur ?
          codegen->GetMurmurHashFunction(expr_values_bytes_per_row) :
          codegen->GetHashFunction(expr_values_bytes_per_row);
      llvm::Value* len = codegen->GetI32Constant(expr_values_bytes_per_row);
      hash_result = builder.CreateCall(
          hash_fn, llvm::ArrayRef<llvm::Value*>({expr_values, len, hash_result}), "hash");
    }
  } else {
    if (var_result_offset > 0) {
      llvm::Function* hash_fn = use_murmur ?
          codegen->GetMurmurHashFunction(var_result_offset) :
          codegen->GetHashFunction(var_result_offset);
      llvm::Value* len = codegen->GetI32Constant(var_result_offset);
      hash_result = builder.CreateCall(
          hash_fn, llvm::ArrayRef<llvm::Value*>({expr_values, len, hash_result}), "hash");
    }

    // Hash string slots