On Fri, 28 Aug 2020 15:58:24 +0200 peterz@xxxxxxxxxxxxx wrote: > On Fri, Aug 28, 2020 at 10:51:13PM +0900, Masami Hiramatsu wrote: > > > OK, schedule function will be the key. I guess the senario is.. > > > > 1) kretporbe replace the return address with kretprobe_trampoline on task1's kernel stack > > 2) the task1 forks task2 before returning to the kretprobe_trampoline > > 3) while copying the process with the kernel stack, task2->kretprobe_instances.first = NULL > > 4) task2 returns to the kretprobe_trampoline > > 5) Bomb! > > > > Hmm, we need to fixup the kernel stack when copying process. > > How would that scenario have been avoided in the old code? Because there > task2 would have a different has and not have found a kretprobe_instance > either. Good question, I think this bug has not been solved in old code too. Let me check. Thanks, -- Masami Hiramatsu <mhiramat@xxxxxxxxxx>
