From the course: Threat Modeling: Denial of Service and Elevation of Privilege
Unlock the full course today
Join today to access over 22,400 courses taught by industry experts or purchase this course individually.
Main forms of corrupt input
From the course: Threat Modeling: Denial of Service and Elevation of Privilege
Main forms of corrupt input
- [Instructor] Let's make this concrete with some specifics. We'll talk about command arguments, including file names, interpreter commands, and processor instructions. Let's say we have a command, open file name. Do we want to allow the thing that sent us input to open every file that we can access? Sometimes that's fine. For example, in an interactive shell. Other times, our code builds filings with input. We might expect that it's going to open a file in the current directory. And the classic attack is to include ../../etc/password in the file name. Depending on what's parsing the file name, you might send it something with a quote mark. In which case, now there's an extra token. Depending on exactly how the arguments to open are parsed. A broader form of this happens with interpreters. That interpreter might be Bash or Python glue code, might be on a server or a serverless implementation. It might be a web browser. In…