On Mon, 13 Jul 2026 13:14:43 +0200, Bartosz Golaszewski <[email protected]> said: > Currently only devicetree systems have their devices' probe ordered > against their suppliers automatically by fw_devlink. Software nodes have > lately been used extensively treewide to describe references to resource > suppliers: most notably, the GPIO subsystem uses it in many places. Now > that the conversion of "dangling" GPIO chip software nodes to using real > links is almost done, it makes sense to ensure no needless probe > deferrals by implementing the add_links() callback from the fwnode > interface. > > This series extends software node support to use fw_devlink and adds test > coverage for the new behaviour. > > Patch 1 adds new kunit helpers that will be used later on for test > cases. > > Patch 2 implements software_node_add_links() modelled on > of_fwnode_add_links(): for every reference property it resolves the > supplier and links to it. There's no allowlist like in DT - a software > node only carries a reference when its author explicitly wants one, so > every reference is an intentional supplier dependency. Graph > "remote-endpoint" references, unregistered supplier software nodes and > self-references are skipped. It also mirrors the device pointer onto a > secondary software node so fw_devlink can find the supplier device, and > purges the fwnode links on release. > > Patches 3 and 5 add the tests: a kunit suite for the add_links() op > itself and GPIO tests for a real-life use-case: a GPIO consumer > referencing its provider via a software node. > > Patch 4 proposes to add myself as a reviewer of software nodes. > > Caveats: a supplier software node must be registered before the consumer > device is added, If the swnode is registered after the consumer was added, > add_links() has already run and set FWNODE_FLAG_LINKS_ADDED, so the late > supplier is missed. Graph/remote- endpoint ordering is left out for now as > well as there are no known users. > > Merging strategy: with an Ack from Kunit maintainers, the entire series > can go through the driver core tree. >
Just a clarification: patch 5/5 will conflict with current gpio/for-next so this one would have to go through the GPIO tree. An immutable branch with commits 1-4 would be great. Bart
