In my thesis, I propose to study two fundamental problems related to the
use of spin locks in multiprocessor real-time systems:
First, if tasks (on different cores) share spin-lock-protected
resources, what is (a safe bound on) the maximum blocking, i.e., the
maximum spin delays, that may arise at runtime? And second, how should
tasks be allocated to processor cores such that all timing
constraints are met despite such blocking?