On Sun, Sep 19, 2010 at 8:36 PM, Kurt M.D John<kurt.md.john () gmail com> wrote:
take a look at the email below. I recently did a pentest and found that a
site was vulnerable to sql injection but it was minimal. The user which runs
the queries has read-only access and the information is public but an sql
injection still spits out the full table nonetheless.
I'm not sure what you mean by that last bit. It could be taken at
least 3 ways, and I'm not sure which you intended:
1. Do you mean that throwing the 1=1 into the form returns a database
error message that reveals the table structure?
2 Do you mean that you were able to query the DBMS instance for table
structure using SQL statements?
3. Do you mean that entering in the 1=1 results in a large number of
rows being returned?
If the first is correct, an error message is not the same as a SQL
injection (though it often is the first sign of an injectable
If the second is correct, I think the rest of the question is already
answered, and all you need to do is provide the SQL statements you
used, along with the table information output, to prove the point.
If the third is correct (which is the one I think you meant), then
given the wildcard hardcoded in the SQL statement, a large number of
rows is to be expected. However, the DBA is incorrect that SQL
injection is not occuring. You have altered the query logic and
appended additional SQL syntax.
Where the query should be :
WHERE PROJECT.FOLIO_NBR LIKE 'some_input_data%'
it has now become:
WHERE PROJECT.FOLIO_NBR LIKE '' or '1'='1%'
These are two completely different SQL statements, and are processed
in entirely different ways by the database engine.
The actual structure of the query has changed, not just the data being
searched for within the database.
The information below
is what the DB Admin sent me in defence of the potential vulnerability. My
question is, since it is vulnerable to sql injection can statements be
mutated to get dangerous results such as privilege escalation, etc. I am not
familiar with sql injection
Since you are testing applications which use databases for security
flaws, you should probably become familiar with SQL injection,
If you need to prove that injection is occurring, try throwing
comments into your injection, and look up using UNION and/or timing
attacks to alter the data being returned. How you do this will vary
based on the DBMS being used. Presuming Microsoft SQL server, comments
could be added by appending two dashes to the end of your injected
Using the example provided in your email, the parameter should become:
' or '1'='1'--
To answer your question of whether this could be used to get dangerous
results, the answer is "it depends". Presuming that an actual
injection was taking place, then it is very likely this could be
leveraged to further an attack (at least to gain more useful
information about the environment), but how much risk is actually
present really varies based (at least) on what the database
environment is, how it is configured, the data it contains, and how
the application is set up to interact with it.
<snipping the remaining stuff>