x86/alternatives: Implement a better poke_int3_handler() completion scheme

Commit:

  285a54ef

 ("x86/alternatives: Sync bp_patching update for avoiding NULL pointer exception")

added an additional text_poke_sync() IPI to text_poke_bp_batch() to
handle the rare case where another CPU is still inside an INT3 handler
while we clear the global state.

Instead of spraying IPIs around, count the active INT3 handlers and
wait for them to go away before proceeding to clear/reuse the data.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Reviewed-by: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>