On 2021-11-24 22:39:34 [-0300], Luis Goncalves wrote: > Sebastian, I just stumbled over the lockdep splat being discussed here > in v5.15.3-rt21. Well, slightly different, but related to the very > same situation: … so we have write_lock_bh(&dev_base_lock) and read_lock(&dev_base_lock) followed by local_bh_disable() (due to spin_lock_bh(&adap->mbox_lock)). It makes sense to only change that one invocation. Let me form a patch. Sebastian
