On Wed, Mar 12, 2025 at 10:31:51AM +0100, David Ahern wrote: > I hope there will be an open mind on get / dump style introspection apis > here. Devices can work support and work within limited subsystem APIs > and also allow the dumping of full essential and relevant contexts for a > device. We have three places to put dumpables like you are talking about, fwctl, netdev and rdma rdma already has a robust dumper for these kinds of objects. netdev drivers have a few options, and people use debugfs there. fwctl is supposed to let you dump the FW side view of the sytem. Do you *really* need another one? It sounds excessive to me. > More specifically, I do not see netdev APIs ever recognizing RDMA > concepts like domains and memory regions. Why not? If netdev is going to provide the same detailed view of the HW state we expect from fwctl it will need to provide ways to dump the actualy underlying HW objects, whatever they may be. It is not like building a netdev driver on top of RDMA is anything new, mlx5 does it as well. > For us, everything is relative to a domain and a region - e.g., > whether a queue is created for a netdev device or an IB QP both use > the same common internal APIs. I would prefer not to use fwctl for > something so basic. It is probably the right answer though. Jason
