When the kernel receives a synchronous external abort with a fault
handler set we can use this to signal that the copy failed.
This has been observed when userspace causes such an abort so a SIGBUS
is raised that causes a coredump to be created. When creating the
coredump another abort is raised from copyin causing the kernel to
panic. Rather than panicking the kernel call the fault handler to
return an error to the caller.
Sponsored by: Arm Ltd