On Thu, Dec 14, 2023 at 3:04 PM Mark Hasemeyer <markhas@xxxxxxxxxxxx> wrote: > > > If a device knows it is wakeup capable, why do you need a property too? > > I'm referencing: > https://www.kernel.org/doc/Documentation/devicetree/bindings/power/wakeup-source.txt > "Nodes that describe devices which has wakeup capability must contain > an "wakeup-source" boolean property." That's probably too strongly worded because wakeup capable devices existed (and still exist) before this binding was created. Powerpc for example doesn't use it. > Currently the driver assumes the device is wake capable without > parsing the device tree, which is an incorrect assumption as wake > capability should not be enabled on some cros_ec systems. > > > I haven't looked closely enough, but it smells like after patch 6, these > > properties would be required for wakeup? That would be an ABI break. > > Agreed. In this case, the driver is a ChromeOS related driver and DTS > is built from source for each OS update. > For more context, I will make sure to CC you (and everyone else) and > include a cover letter in the next series version. Please explain in the patches with an ABI break why it doesn't matter. Rob
