FAU_GEN.1 mandatory events for FCS_RBG.1

[woodbe]

If the TOE us invoking platform-provided functionality for generating random bits, it isn't clear how the TOE would be able to log any output from an error. It is unlikely to even know there is an error in that case.

I think this needs to be written like the FCS_CKM.1 event with a selection that can be tailored to the specific implementation.

[jfisherbah]

In this case the 'failure of the randomization process' is probably covered implicitly by FPT_TST.1 because it could be argued that a self-test failure is the only way that such a failure would be evident and FPT_TST.1.1 specifies "TSF DRBG" as the thing that is being self-tested. So in this case it's probably appropriate to remove the auditable event from FCS_RBG.1 since it's not directly related to the SFR itself.

However this would raise a similar question that if the TSF invokes a platform-provided DRBG, there is no "TSF DRBG" to self-test. In this situation would it be more appropriate to move FPT_TST.1 to selection-based, with the trigger being that the DRBG is implemented within the TOE boundary?