Hi Bart Van Assche and Jens Axboe Sorry to disturb you. Would you have a plan When will these patch sets be merged into the mainline? Thanks! -----邮件原件----- 发件人: Bart Van Assche <bvanassche@xxxxxxx> 发送时间: 2024年5月17日 5:28 收件人: YangYang <yang.yang@xxxxxxxx> 抄送: linux-block@xxxxxxxxxxxxxxx; Christoph Hellwig <hch@xxxxxx>; Damien Le Moal <dlemoal@xxxxxxxxxx>; 牛志国 (Zhiguo Niu) <Zhiguo.Niu@xxxxxxxxxx>; Jens Axboe <axboe@xxxxxxxxx> 主题: Re: [PATCH v2 2/2] block/mq-deadline: Fix the tag reservation code 注意: 这封邮件来自于外部。除非你确定邮件内容安全，否则不要点击任何链接和附件。 CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. On 5/16/24 02:14, YangYang wrote: >> @@ -513,9 +527,9 @@ static void dd_depth_updated(struct blk_mq_hw_ctx >> *hctx) >> struct deadline_data *dd = q->elevator->elevator_data; >> struct blk_mq_tags *tags = hctx->sched_tags; >> - dd->async_depth = max(1UL, 3 * q->nr_requests / 4); >> + dd->async_depth = q->nr_requests; >> - sbitmap_queue_min_shallow_depth(&tags->bitmap_tags, >> dd->async_depth); >> + sbitmap_queue_min_shallow_depth(&tags->bitmap_tags, 1); > > If sbq->min_shallow_depth is set to 1, sbq->wake_batch will also be > set to 1. I guess this may result in batch wakeup not working as expected. The value of the sbq->min_shallow_depth parameter may affect performance but does not affect correctness. See also the comment above the sbitmap_queue_min_shallow_depth() declaration. Is this sufficient to address your concern? Thanks, Bart.
