Is Perl's taint mode useful?

Question

perl -T

Do you use it? Does it help you finding security holes in your Perl scripts?

Answer 1

More than that :) it stops your security issues before they become one. It is not a security silver bullet of course... we used to use it (a few years back when I was involved in Perl projects) in any script that was exposed externally (i.e. any mod_perl app) and we found it very useful and made it our policy. It does a few checks and it is handy.. (anything makes things automated)

Perl Security - perlsec recommends it strongly too:

This flag [Taint mode] is strongly suggested for server programs and any program run on behalf of someone else, such as a CGI script. Once taint mode is on, it's on for the remainder of your script.

Answer 2

Most definitely!

$ echo '`rm -rf /`' | perl -Te 'eval while <>'
Insecure dependency in eval while running with -T switch at -e line 1, <> line 1.

Answer 3

The "Secure Programming Techniques" chapter of Mastering Perl is almost completely devoted to taint checking and how you should use it.

Many people will tell you it protects you, but they subtly lie about that. It's a developer tool that helps you find some (only some) spots in your code where you need to be careful. It's not going to solve all of your security problems.

Answer 4

I think taint mode would work best when new code is being developed that everyone is familiar with.

If you have someone else's code that is poorly written, and you run it in Taint mode -- perl will die rather than perform what by the tainting rules are 'unsafe' operations.

In taint mode perl some holes are patched but not all. system("$unfiltered_user_input") will die but Perl could still write $unfiltered_user_input data to a file with a fixed name (because printing tainted data is considered 'safe') and then execute that file with system(). But nothing can check everything.

There's a tradeoff there for using it on legacy apps. When Perl finds an unsafe operation on tainted data it will die -- which means someone must go in and decide what it means to untaint the data, what regexp are needed, before the application will be reliable again.

Some people would prefer insecure, reliable, low cost (for now) to -- secure, broken, need to find the developers. Not that thats good in the long run... but it is not unusual.

Answer 5

Yes, taint mode is useful for all the reasons mentioned above.

One place that you may not consider tainted data is when interacting with a database. Fortunately, DBI has support for stopping tainted data from getting into your database, and it treats data coming from your database as being tainted so that you can't do anything unsafe with it. You have to specifically turn on the options for this; they're off by default. See the DBI docs for more.