Messages in this thread | | | Date | Wed, 29 Oct 2014 12:36:55 +0100 | From | Peter Zijlstra <> | Subject | Re: [PATCH 00/11] nested sleeps, fixes and debug infrastructure |
| |
On Wed, Oct 29, 2014 at 12:31:03PM +0100, Peter Zijlstra wrote: > On Wed, Oct 29, 2014 at 10:35:03AM +0100, Peter Zijlstra wrote: > > On Wed, Oct 29, 2014 at 01:00:56AM +0100, Oleg Nesterov wrote: > > > > But what if it calls try_to_freeze() and another thread (which should > > > be frozen too) sleeps in kthread_stop() ? > > > > Fair point indeed. Now I had a look at __refrigerator() and is there any > > reason we should not remove that .check_kthr_stop argument and replace > > it with an unconditional (current->flags & PF_KTHREAD) ? > > Hmm, doesn't kthread_freezable_should_stop() suffer the same problem? It > should not refrigerate when should_stop.
Oh never you mind, i've got my head on backwards it seems, it uses __refrigerator(.check_kthr_stop = true), which does another should_stop test.
| |