blocking SIGSEGV and triggering a segment violation produces an all CPU consuming process
If a process blocks (or ignores?) SIGSEGV (or SIGFPE, SIGILL, SIGBUS, etc.) and
then triggers the associated signal without having a handler installed, the
process/thread will be stuck on the triggering instruction without any hope.
I think the kernel should detect such a situation (i.e. trap leading to
blocked/ignored signal without handler) and should kill the process.