Hello,
for a functional safety project we are currently trying to determine/evaluate the possible dangers caused by the usages of floats (and float operations/calculations). As far as we can tell, there are many ways to archive for example undefined behaviors when it comes to floats.
For example:
- division by zero,
- overflow,
- underflow,
- trap representations read by an lvalue (from C18 Standard, Annex J.2),
- demotion of one real float to another producing a value outside of the range that can be represented (from C18 Standard, Annex J.2)
- an argument to a floating-point classification or comparison macro is not real floating type (from C18 Standard, Annex J.2)
- etc.
So, we were wondering: What is the worst that can happen when using floats or float operations/calculations?
- Can a memory corruption happen (read-only memory and/or RAM)?
- Can the program code or/and program flow be manipulated?
- Or what other dangerous influences can the usage of floats have?
Kind regards
Sebastian Ahrens
