On Fri, May 31, 2019 at 9:35 PM Stephen Smalley <sds@xxxxxxxxxxxxx> wrote: > > + entry->lsm[i].rule = rule_new; > > Doesn't this still leak the old entry->lsm[i].rule? Argh, clearly got a wrong understanding from different part of the code. Will fix. > Also, I don't think you can just mutate entry like this under RCU. Yeah, it's definitely not the politically correct way of doing it. Let's rework the entire list then, I will post another draft. It will become somewhat more intrusive :-( -- Janne
