x86: Issue at least one memory barrier in stop_machine_text_poke()

Fix stop_machine_text_poke() to issue smp_mb() before exiting
waiting loop, and use cpu_relax() for waiting.

Changes in v2:
 - Don't use ACCESS_ONCE().
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: systemtap <systemtap@sources.redhat.com>
Cc: DLE <dle-develop@lists.sourceforge.net>
Cc: Jason Baron <jbaron@redhat.com>
LKML-Reference: <20100304033850.3819.74590.stgit@localhost6.localdomain6>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

arch/x86/kernel/alternative.c

@@ -595,8 +595,8 @@ static int __kprobes stop_machine_text_poke(void *data)
 		wrote_text = 1;
 	} else {
 		while (!wrote_text)
-			smp_rmb();
-		sync_core();
+			cpu_relax();
+		smp_mb();	/* Load wrote_text before following execution */
 	}
 
 	flush_icache_range((unsigned long)tpp->addr,