On Fri, Mar 5, 2021 at 9:13 AM Andy Shevchenko
[off-list ref] wrote:

On Thu, Mar 4, 2021 at 11:40 PM Hector Martin [off-list ref] wrote:

quoted

This implements the 'nonposted-mmio' and 'posted-mmio' boolean
properties. Placing these properties in a bus marks all child devices as
requiring non-posted or posted MMIO mappings. If no such properties are
found, the default is posted MMIO.

of_mmio_is_nonposted() performs the tree walking to determine if a given
device has requested non-posted MMIO.

of_address_to_resource() uses this to set the IORESOURCE_MEM_NONPOSTED
flag on resources that require non-posted MMIO.

of_iomap() and of_io_request_and_map() then use this flag to pick the
correct ioremap() variant.

This mechanism is currently restricted to Apple ARM platforms, as an
optimization.

...

quoted

@@ -896,7 +899,10 @@ void __iomem *of_iomap(struct device_node *np, int index)
        if (of_address_to_resource(np, index, &res))
                return NULL;

-       return ioremap(res.start, resource_size(&res));
+       if (res.flags & IORESOURCE_MEM_NONPOSTED)
+               return ioremap_np(res.start, resource_size(&res));
+       else
+               return ioremap(res.start, resource_size(&res));

This doesn't sound right. Why _np is so exceptional? Why don't we have
other flavours (it also rings a bell to my previous comment that the
flag in ioresource is not in the right place)?

...

quoted

+       if (res.flags & IORESOURCE_MEM_NONPOSTED)
+               mem = ioremap_np(res.start, resource_size(&res));
+       else
+               mem = ioremap(res.start, resource_size(&res));
+

Ditto.

...

quoted

+       while (node) {
+               if (!of_property_read_bool(node, "ranges")) {
+                       break;
+               } else if (of_property_read_bool(node, "nonposted-mmio")) {
+                       of_node_put(node);
+                       return true;
+               } else if (of_property_read_bool(node, "posted-mmio")) {
+                       break;
+               }
+               parent = of_get_parent(node);
+               of_node_put(node);
+               node = parent;
+       }

I believe above can be slightly optimized. Don't we have helpers to
traverse to all parents?

We don't. I only found a handful of cases mostly in arch/powerpc.
Given that and this series is big enough already, I don't think we
need a helper as part of it. But patches welcome.

Rob