[RFT PATCH v3 12/27] of/address: Add infrastructure to declare MMIO as non-posted

[Arnd Bergmann]

On Thu, Mar 11, 2021 at 1:11 PM Hector Martin <marcan at marcan.st> wrote:
> On 11/03/2021 18.12, Arnd Bergmann wrote:
> > On Wed, Mar 10, 2021 at 6:01 PM Rob Herring <robh at kernel.org> wrote:
> >> On Wed, Mar 10, 2021 at 1:27 AM Hector Martin <marcan at marcan.st> wrote:
> >>> Works for me; then let's just make it non-recursive.
> >>>
> >>> Do you think we can get rid of the Apple-only optimization if we do
> >>> this? It would mean only looking at the parent during address
> >>> resolution, not recursing all the way to the top, so presumably the
> >>> performance impact would be quite minimal.
> >
> > Works for me.
>
> Incidentally, even though it would now be unused, I'd like to keep the
> apple,arm-platform compatible at this point; we've already been pretty
> close to a use case for it, and I don't want to have to fall back to a
> list of SoC compatibles if we ever need another quirk for all Apple ARM
> SoCs (or break backwards compat). It doesn't really hurt to have it in
> the binding and devicetrees, right?

Yes, keeping the compatible string is a good idea regardless.

> >> Yeah, that should be fine. I'd keep an IS_ENABLED() config check
> >> though. Then I'll also know if anyone else needs this.
> >
> > Ok, makes sense.
> >
> > Conceptually, I'd like to then see a check that verifies that the
> > property is only set for nodes whose parent also has it set, since
> > that is how AXI defines it: A bus can wait for the ack from its
> > child node, or it can acknowledge the write to its parent early.
> > However, this breaks down as soon as a bus does the early ack:
> > all its children by definition use posted writes (as seen by the
> > CPU), even if they wait for stores that come from other masters.
> >
> > Does this make sense to you?
>
> Makes sense. This shouldn't really be something the kernel concerns
> itself with at runtime, just something for the dts linting, right?
>
> I assume this isn't representable in json-schema, so it would presumably
> need some ad-hoc validation code.

Agreed, having a check in either dtc or expressed in the json scheme
is better than a runtime check. I assume Rob would know how to best
add such a check.

     Arnd