RE: [PATCH 0/3] blk-mq: improve IO perf in case of none io sched

[Kashyap Desai <kashyap.desai@broadcom.com>]

> Right.
>
> Kashyap, could you test the following patch?
>
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index 2f20c9e3efda..7d972b1c3153 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -1567,7 +1567,7 @@ void blk_mq_insert_requests(struct blk_mq_hw_ctx
> *hctx, struct blk_mq_ctx *ctx,
>  	}
>
>  	spin_lock(&ctx->lock);
> -	list_splice_tail(list, &ctx->rq_list);
> +	list_splice_tail_init(list, &ctx->rq_list);
>  	blk_mq_hctx_mark_pending(hctx, ctx);
>  	spin_unlock(&ctx->lock);

I tested above mentioned change and kernel panic as posted in this
discussion is resolved.

Kashyap

>  }
>
> Thanks,
> Ming