On 5/26/21 8:33 PM, Damien Le Moal wrote:
> On 2021/05/27 10:02, Bart Van Assche wrote:
>> For interactive workloads it is important that synchronous requests are
>> not delayed. Hence reserve 25% of tags for synchronous requests. This patch
>
> s/tags/scheduler tags
>
> to be clear that we are not talking about the device tags. Same in the patch
> title may be.
OK.
>> +static void dd_limit_depth(unsigned int op, struct blk_mq_alloc_data *data)
>
> Similarly as you did in patch 1, may be add a comment about this operation and
> when it is called ?
Will do.
>> +static void dd_depth_updated(struct blk_mq_hw_ctx *hctx)
>> +{
>> + struct request_queue *q = hctx->queue;
>> + struct deadline_data *dd = q->elevator->elevator_data;
>> + struct blk_mq_tags *tags = hctx->sched_tags;
>> +
>> + dd->async_depth = 3 * q->nr_requests / 4;
>
> I think that nr_requests is always at least 2, but it may be good to have a
> sanity check here that we do not end up with async_depth == 0, no ?
OK, I will add a check.
Thanks,
Bart.
