On Wed, 2024-05-01 at 07:39 -0500, Rob Herring wrote:

On Tue, Apr 30, 2024 at 7:40 PM Andrew Jeffery
[off-list ref] wrote:

quoted

On Tue, 2024-04-30 at 12:25 -0500, Rob Herring (Arm) wrote:

quoted

The use of 'oneOf' to include 1 of 3 possible child node schemas results
in error messages containing the actual error message(s) for the correct
SoC buried in the tons of error messages from the 2 schemas that don't
apply. It also causes the pinctrl schema to be applied twice as it will
be applied when the compatible matches.

All that's really needed in the parent schema is to ensure one of the
possible compatible strings is present in the pinctrl node so that its
schema will be applied separately.

Thanks, I think it improves the readability of intent in the binding as
well.

To understand the impact better I grabbed the patch and diffed the
output of `make CHECK_DTBS=y aspeed/aspeed-ast2600-evb.dtb` before and
after applying it, but there was no significant difference in output.
Should that not demonstrate the errors being cleaned up? If not, what
should?

Try it on one of the new boards posted in the last 1-2 days. It showed
up on my testing dtbs_check on patches. I didn't send a report because
there was so much noise in it.

I tried with aspeed/aspeed-bmc-ibm-blueridge.dtb and yeah, it does
clean up a lot of barf. Nice. Thanks.

Reviewed-by: Andrew Jeffery <andrew@codeconstruct.com.au>

[1]: https://lore.kernel.org/lkml/20240429210131.373487-14-eajames@linux.ibm.com/ (local)

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel