# Re: \[PATCH net-next v2\] net: opt loopback into instance locking

- From: Matthieu Baerts \<matttbe@kernel.org\>
- Date: 2026-09-22 07:11:09 
- List: netdev
- Page: https://ratatoskr.run/netdev/2026/09/17619942
- Thread: https://ratatoskr.run/netdev/2026/09/17617285/t.md

```text
Hi Jakub,

On 21/09/2026 21:04, Jakub Kicinski wrote:
> netdev netlink iterates over devices when dumping queues/NAPIs/qstats
> etc. and takes rtnl_lock or instance lock depending on the underlying
> device. Currently even if all the "real" devices in the system are ops
> locked we still have to take rtnl_lock for lo (only to find out
> that it doesn't even have queues or NAPIs to report).
> Opt loopback into having control path under the ops lock.
> 
> This is a pretty obvious thing to do, I've held back this
> patch because we used to only apply ops locking on physical devices.
> netkit queue leasing made a precedent for (far more complex)
> SW devices enabling ops locking. Now adding it to lo should
> not create much new bug surface.
> 
> I considered an alternative of adding a "predicate" to the iteration
> primitive so that we can skip the devices which obviously don't
> support given API (eg qstat) without any locking. But it's more
> LoC and real_num_.x_queues is not currently WRITE_ONCE()ed so
> it doesn't work too well for queues.

Thank you for this! But apparently, it looks like it is still causing
some issues with the BPF selftests for some arch, e.g.

https://github.com/kernel-patches/bpf/actions/runs/35693264483/job/106636983010

Cheers,
Matt
---
(I don't know if I have the rights to ask for ↓)
pw-bot: cr
```
