2

I am watching a training video called "Creating Secure J2EE Code". The video makes this statement:

Perform all validation in a secure context on the server. Do not rely on client-side validation as a security mechanism. Validate all application input, including user input, API return values, environment variables, configuration files, and other applications.

Validating environment variables and config files is a new concept to me. Can you please provide a high level overview of what this might look like?

My thoughts are to treat both Environment Variables and Config files similar to user input. I would:

  • Sanitize it,
  • Test Length (Min & Max),
  • Test Range,
  • Test format
  • Test type (int, decimal, char, etc)

I am assuming config files would contain Key/Value pairs. Im not certain how to handle a configuration file that may contain any other format.

Am I on the right track?

  • Normally, I only recommend validating data from untrusted sources unless it is easy to implement. For example, if your web app loads data from a business partner I would validate that if your app is going to process/display it. If it is loading a config file that only trusted admins can edit then there's less risk (depending on your threat model). It's not a bad idea to validate all data to reduce errors, but I'd focus your energy on validating untrusted data first. – PwdRsch Jun 30 '14 at 14:46
3

(this should probably be a comment but i's a bit verbose)

Validating environment variables and config files is a new concept to me

At first this struck me as odd - then I remembered that not alJ2EE applications run in a controlled environment. It's a question of trust. I deal with web applications where I'd put more trust in the config of the environment than in the code itself - here that statement makes no sense. OTOH if I were writing an app to be run on desktop PCs then I'd be a bit suspicious of anything fetched from the local system.

I would....sanitize it

No.

Validate inputs (i.e. accept or reject them in their entirety)

Sanitize outputs (convert them to a representation appropriate to the medium they are going into).

Sanitizing inputs is a really bad idea. You don't know where the data is going until your code looks at it. If it can arrive in your code in a state where it can do something unexpected (code injection, buffer overflow etc) then its too late to fix the problem.

  • This is great insight. @PwdRsch, thank you for commenting as well. – medokr Jun 30 '14 at 18:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.