原代码如下

private Symbol translateCallExpr(TAProgram program, ASTNode node, SymbolTable symbolTable) throws ParseException {
        //函数名称
        var factor = node.getChild(0);



        var list = new ArrayList<TAInstruction>();
        //函数入参
        for(int i = 1; i < node.getChildren().size(); i++) {
            var expr = node.getChildren().get(i);
            //由于参数可以是表达式，所以通过translateExpr进行翻译
            var addr = translateExpr(program, expr, symbolTable);
            list.add(new TAInstruction(TAInstructionType.PARAM, null, null, addr, i - 1));
        }

        //将arg2设置为offset
        for(var instruction : list) {
            instruction.setArg2( symbolTable.localSize() + (int)instruction.getArg2() + 2);
            program.add(instruction);
        }

        //函数返回值
        var returnValue = symbolTable.createVariable();
        //函数返回地址
        var funcAddr = symbolTable.cloneFromSymbolTree(factor.getLexeme(), 0);
        if(funcAddr == null) {
            throw new ParseException("function " + factor.getLexeme().getValue() + " not found");
        }
        //函数压栈，消耗栈空间symbolTable.localSize()个单位
        program.add(new TAInstruction(TAInstructionType.SP, null, null,
                -symbolTable.localSize(), null));
        program.add(new TAInstruction(TAInstructionType.CALL, null, null, funcAddr, null));
        //函数出栈，恢复栈空间
        program.add(new TAInstruction(TAInstructionType.SP, null, null,
            symbolTable.localSize(),null));
        return returnValue;
    }

为什么在设置入参偏移量的时候会+2，视频中好像没解释？

   //将arg2设置为offset
        for(var instruction : list) {
            instruction.setArg2( symbolTable.localSize() + (int)instruction.getArg2() + 2);
            program.add(instruction);
        }