On Thu, Jul 05, 2018 at 09:58:48AM -0700, Paul E. McKenney wrote: > On Thu, Jul 05, 2018 at 05:39:06PM +0200, Andrea Parri wrote: > > > > At any rate, it looks like instead of strengthening the relation, I > > > > should write a patch that removes it entirely. I also will add new, > > > > stronger relations for use with locking, essentially making spin_lock > > > > and spin_unlock be RCsc. > > > > > > Only in the presence of smp_mb__after_unlock_lock() or > > > smp_mb__after_spinlock(), correct? Or am I confused about RCsc? > > > > There are at least two definitions of RCsc: one as documented in the header > > comment for smp_mb__after_spinlock() or rather in the patch under review..., > > one as processor architects used to intend it. ;-) > > Searching isn't working for me all that well this morning, so could you > please send me a pointer to that patch?
Sorry, I meant in _this patch_: "RCsc" as ordering everything except for W -> R, without the barriers above (_informally, the current LKMM misses the W -> W order only). Andrea > > Thanx, Paul >
