Fault injection is a powerful technique for attacking digital systems. Software developers have to take into account hardware fault effects when system security is a concern. Software fault models have been developed in an attempt to predict these faults. However, these models are often designed independently of any hardware consideration and thus raise the problem of realism. The generality of these models often cannot account for the specificities of each architecture. As a consequence, software countermeasures based on such software fault models do not guarantee an effective protection against fault attacks. Processor microarchitecture should be precisely analysed to better understand faulty behaviours. A crosslayer approach can then be developed, using conjointly hardware and software characteristics to design stronger software countermeasures with reasonable overheads. To illustrate this assumption, this paper shows actual faulty behaviours observed in a RISC-V processor RTL simulation, and shows that they can bypass countermeasures designed to protect against faults predicted by typical software fault models.