Re: [PATCH 01/20] asm-generic/mmiowb: Add generic implementation of mmiowb() tracking
From: Nicholas Piggin
Date: Sun Mar 03 2019 - 05:05:56 EST
Linus Torvalds's on March 3, 2019 2:29 pm:
> On Sat, Mar 2, 2019, 19:34 Nicholas Piggin <npiggin@xxxxxxxxx> wrote:
>> It doesn't have to be done all at once with this series, obviously this
>> is a big improvement on its own. But why perpetuate the nomenclature
>> and concept for new code added now?
> What nomenclature?
> Nobody will be using mmiowb(). That's the whole point of the patch series.
> It's now an entirely internal name, and nobody cares.
Why even bother with it at all, "internal" or not? Just get rid of
mmiowb, the concept is obsolete.
> And none of this has anything to do with wmb(), since it's about IO being
> ordered across cpu's by spin locks, not by barriers.
> So I'm not seeing what you're arguing about.
Pretend ia64 doesn't exist for a minute. Now the regular mb/wmb barriers
orders IO across CPUs with respect to their cacheable accesses.
Regardless of whether that cacheable access is a spin lock, a bit lock,
an atomic, a mutex... This is how it was before mmiowb came along.
Nothing wrong with this series to make spinlocks order mmio, but why
call it mmiowb? Another patch could rename ia64's mmiowb and then the
name can be removed from the tree completely.