Hi Sagi, On Fri, Nov 15, 2019 at 02:38:44PM -0800, Sagi Grimberg wrote: > > > Hi, > > Hey Ming, > > > Use blk_mq_alloc_request() for allocating NVMe loop, fc, rdma and tcp's > > connect request, and selecting transport queue runtime for connect > > request. > > > > Then kill blk_mq_alloc_request_hctx(). > > Is it really so wrong to have an API to allocate a tag that belongs to > a specific queue? Why must the tags allocation always correlate to the > running cpu? Its true that NVMe is the only consumer of this at the > moment, but does this mean that the interface should be removed because > it has one (or rather four) consumer(s)? Now blk-mq takes a static queue mapping between CPU and hw queues, given CPU hotplug may happen any time, so the specified hw queue may become inactive any time. Queue mapping from CPU to hw queue is one core model of blk-mq which relies a lot on the fact that hw queue active or not depends on request's submission CPU. And we always can retrieve one active hw queue if there is at least one online CPU. IO request is always mapped to the proper hw queue via the submission CPU and queue type. So blk_mq_alloc_request_hctx() is really weird from the above blk-mq's model. Actually the 4 consumer is just one single type of usage for submitting connect command, seems no one explain this special usage before. And the driver can handle well enough without this interface just like this patch, can't it? > > I would instead suggest to simply remove the constraint that > blk_mq_alloc_request_hctx() will fail if the first cpu in the mask > is not on the cpu_online_mask.. The caller of this would know and > be able to handle it. Of course, this usage is very special, which is different with normal IO or passthrough request. The caller of this still needs to rely on blk-mq for dispatching this request: 1) blk-mq needs to run hw queue in round-robin style, and different CPU is selected from active CPU masks for running the hw queue. 2) Most of blk-mq drivers have switched to managed IRQ, which may be shutdown even though there is still in-flight requests not completed on the hw queue. We need to fix this issue. One solution is to free the request and remap the bios into proper active hw queue according to the new submission CPU. 3) warning will be caused when dispatching one request on inactive hw queue If we are going to support this special usage, lots of blk-mq needs to be changed for covering the so special case. > > To me it feels like this approach is fundamentally wrong. IMO, having > the driver select a different queue than the tag naturally belongs to > feels like a backwards design. The root cause is that the connection command needs to be submitted via one specified queue, which is fundamentally not compatible with blk-mq's queue mapping model. Given the connect command is so special for nvme, I'd suggest to let driver handle it completely, since blk-mq is supposed to serve generic IO function. Thanks, Ming
