On Mon, May 18, 2020 at 03:16:34PM +0200, Christoph Hellwig wrote: > On Mon, May 18, 2020 at 07:54:54PM +0800, Ming Lei wrote: > > > > I guess I misunderstood your point, sorry for that. > > > > The requirement is just that the request needs to be allocated on one online > > CPU after INACTIVE is set, and we can use a workqueue to do that. > > I've looked over the code again, and I'm really not sure why we need that. > Presumable the CPU hotplug code ensures tasks don't get schedule on the > CPU running the shutdown state machine, so if we just do a retry of the percpu kthread still can be scheduled on the cpu to be online, see is_cpu_allowed(). And bound wq has been used widely in fs code. > tag allocation we can assume we are on a different CPU now (Thomas, > correct me if that assumption is wrong). So I think we can do entirely > with the preempt_disable and the smp calls. Something like this branch, > which has only survived basic testing (the last patch is the pimary > interesting one, plus maybe the last 3 before that): > > git://git.infradead.org/users/hch/block.git blk-mq-hotplug.2 > > Gitweb: > > http://git.infradead.org/users/hch/block.git/shortlog/refs/heads/blk-mq-hotplug.2 > After preempt_disable() is removed, the INACTIVE bit is set in the CPU to be offline, and the bit can be read on all other CPUs, so other CPUs may not get synced with the INACTIVE bit. Thanks, Ming
