On Thu, Apr 01, 2021 at 08:14:39PM CDT, Andrew Jeffery wrote:

On Fri, 2 Apr 2021, at 11:17, Zev Weiss wrote:

quoted

These correspond to the existing lpc_address, sirq, and sirq_polarity
sysfs attributes; the second element of aspeed,sirq provides a
replacement for the deprecated aspeed,sirq-polarity-sense property.

Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
---
 .../devicetree/bindings/serial/8250.yaml      | 27 ++++++++++++++++---
 1 file changed, 24 insertions(+), 3 deletions(-)

diff --git a/Documentation/devicetree/bindings/serial/8250.yaml

b/Documentation/devicetree/bindings/serial/8250.yaml
index 491b9297432d..a6e01f9b745f 100644

--- a/Documentation/devicetree/bindings/serial/8250.yaml
+++ b/Documentation/devicetree/bindings/serial/8250.yaml

@@ -12,8 +12,13 @@ maintainers:
 allOf:
   - $ref: /schemas/serial.yaml#
   - if:
-      required:
-        - aspeed,sirq-polarity-sense
+      anyOf:
+        - required:
+            - aspeed,lpc-address

Why not aspeed,lpc-io-reg like the KCS binding?

There are some things we can do to improve it, but we shouldn't go and invent something different. I prefer aspeed,lpc-io-reg because it's name derives from the generic 'reg' property as does it's behaviour (if you assume a related `#size-cells = 0`).

quoted

+        - required:
+            - aspeed,sirq

Why not aspeed,lpc-interrupts like the KCS binding?

The generic IRQ property is 'interrupts', so like aspeed,lpc-io-reg the interrupts proposal for KCS follows in name and form. I'm hiding it behind the aspeed vendor prefix for now while I'm not satisfied that I understand the requirements of non-aspeed parts. Similarly, I added the lpc prefix because we don't tend to describe the host devicetree in the BMC devicetree (and so there's no parent interrupt controller that we can reference via a phandle) and we need a way to differentiate from the local interrupts property.

I don't see a reason for either of them to differ from what we already have for KCS, and I don't see any reason to continue the sysfs naming scheme in the binding.

Ah, OK -- I was aiming for consistency with the existing vuart sysfs 
attributes, but if that's not a worthwhile concern I'm fine with 
aspeed,lpc-io-reg & aspeed,lpc-interrupts.


Zev