Hi Ido, On Sun, 28 Jun 2026 20:49:00 +0300, Ido Schimmel wrote: > I don't think anything needs to change. AFAICT, br_mst_info_size() is > only reachable via the get_link_af_size() callback and > rtnl_link_get_af_size() always invokes it from an RCU read-side critical > section. > > Did you see a splat with CONFIG_PROVE_RCU_LIST? Thanks for the clarification.
No, I did not see an in-kernel splat from the actual rtnetlink path. The warning I had was from the reviewed out-of-tree reproducer used for our CONFIG_PROVE_RCU_LIST triage, not from a real bridge netlink call path. Given your explanation that rtnl_link_get_af_size() already invokes this from an RCU read-side critical section, I will treat this as an intended contract and I will not pursue a patch for it based on the current evidence. Thanks, Runyu
