On 3/27/22 19:20, Bart Van Assche wrote:
Thanks, this helps a lot. can_queue = 1 as I was suspecting. In a quick test I noticed that I/O is about 10x slower for queue depth 1 and the v5.17 mq-deadline scheduler. I will take a closer look at this tomorrow.
An update: I can reproduce the slowness with (a) the patch at the start of this thread reverted and (b) both the BFQ and the Kyber I/O schedulers. So it seems like a regression has been introduced related to handling of queue depth 1. I will try to bisect this. Bart.
