On 06/04/20 21:58, Eric Blake wrote: >> >> +For the purposes of pairing, ``smp_read_barrier_depends()`` and >> ``smp_rmb()`` >> +both count as read barriers. A read barrier shall pair with a write >> +barrier or a full barrier; a write barrier shall pair with a read > > 'shall' is awkward (if this is not a formal RFC-style requirement), > better for colloquial English is 'must' or 'should' (twice)
Yes, it is a formal requirement. If you don't pair barriers, you don't get any synchronization between the threads. Thanks for the edits! I have included them, except when the text is going to be removed by the next patch. Paolo
