On 06/12/2025 09:45, Bartosz Golaszewski wrote:
And the difference here is that there is no devlink between reset-gpio and its consumers. We need to first agree how to add it.
So you mean that between every other reset consumer and reset provider there is a devlink? And here there is no devlink?
Effectively: yes, but only because reset is OF-only (as of commit 8bffbfdc01df ("reset: remove legacy reset lookup code")) so device links are created from device-tree. If you look at any device using reset-gpio in sysfs, you'll see a supplier:gpiochipX entry but no entry for the reset. So the answer is: yes, there's no devlink between the reset-gpio provider and its consumers unless we create them explicitly, which we should eventually do but I need to rethink the patch because we should possibly also remove the devlink between the gpiochip and the reset consumer as well.
Of course, here we could mention a different story: reset seems like one of these subsystems suffering from object lifetime issues: if you remove the supplier and the consumer tries to use the reset, you'll crash because of the dangling rstc->rcdev pointer. That should be addressed as well eventually.
Yeah, I get it. We talked in person, so unless you plan to rework the patch to fix it differently:
Reviewed-by: Krzysztof Kozlowski krzysztof.kozlowski@oss.qualcomm.com
Best regards, Krzysztof