Re: IBM i web security (was SSL public facing web site)

There is an easy way, use PHP, without checking the request. :-)

In RPG you can use "static SQL" or dinamic. The static will be analized by the pre-compiler and translated to requests to the QSL engin. If a char field has a SQL instruction, it will be treates as data, not executed.

With dinamic SQL you build a string and pass it to the SQL when you run the program. If the string is buildt using data suplied by the user, it is vulnerable to "SQL injection"

Bradley Stone wrote:

Does anyone have an example of manipulating data, running programs,
compromising security, etc on web server running on an IBM i, or is it
purely speculative (or improper setup of the Apache config file, network
mapping, programming, etc?)

I know on other systems there are SQL exploits where you can try to pass
SQL statements into form fields but in my testing on applications (that
I've created that use SQL) I couldn't cause this issue. I've even read
recently where there are loopholes in things like Ruby on Rails, but I'm
mainly talking about CGI programs written with RPG and the pbApache server
running on the IBM i.

I'm not asking for strawman arguments. We can assume that the config file
is set up properly, the port mapping is correct, etc. I would even be
happy to provide a sample apache config file. Port 80 and/or 443 are the
only ports mapped to the i, etc.

I guess what I'm looking for, as are others I'm sure, are some examples on
how things could go wrong instead of simple speculation.

(sounds like a Friday question! hehe...)

Brad
www.bvstools.com