On Wed, May 13, 2020 at 01:59:32PM +0200, Christoph Hellwig wrote: > > + if (unlikely(direct_issue && rq->mq_ctx->cpu != raw_smp_processor_id())) > > + smp_mb(); > > + else > > + barrier(); > > Independ of the fact that I still think this is horrible and hacky, Why? I believe it has been documented clearly. Or can you share better idea? > I also still don't see what the barrier() is trying to help with. It is a compiler barrier, the two OPs are still to be ordered in case of single CPU. Thanks, Ming
