block-mq: fix hung due to too much warning log

When doing cpu online/offline, sometimes this warning will be triggered, it
is harmless, the io will be dispatched. Just sometimes it warns too much that
could hung the whole system. So ratelimit it.

Orabug: 30544814
Signed-off-by: Junxiao Bi <[email protected]>
Reviewed-by: Joe Jin <[email protected]>
Signed-off-by: Somasundaram Krishnasamy <[email protected]>

Author: biger410

block/blk-mq.c

@@ -1212,7 +1212,7 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
 	 *   handle dispatched requests to this hctx
 	 */
 	if (!cpumask_test_cpu(raw_smp_processor_id(), hctx->cpumask) &&
-		cpu_online(hctx->next_cpu)) {
+		cpu_online(hctx->next_cpu) && printk_ratelimit()) {
 		printk(KERN_WARNING "run queue from wrong CPU %d, hctx %s\n",
 			raw_smp_processor_id(),
 			cpumask_empty(hctx->cpumask) ? "inactive": "active");