On Sun, 2021-09-26 at 17:10 +0200, Thomas Gleixner wrote: > > That sequence count is protected by brmctx->br->multicast_lock, but > seqcount_init() does not associate the lock. > > Untested below should cure it properly. Yup, it did. Hrmph, there were even free samples in older trees.. that live right in front of me. Bugger... oh well, bring it on trivial seqcount bugs, ya barely scratched me! <shakes fist> :) -Mike