blktrace: don't increase blk_probes_ref if failed to setup blk trace

do_blk_trace_setup() may return EBUSY, but the current code
doesn't decrease blk_probes_ref in this case.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <49C2F5FF.80002@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

kernel/trace/blktrace.c

@@ -468,16 +468,16 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		blk_register_tracepoints();
-
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
 	if (old_bt) {
 		(void) xchg(&q->blk_trace, old_bt);
 		goto err;
 	}
 
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
+
 	return 0;
 err:
 	if (bt) {