On Tue, Oct 3, 2023 at 4:16 AM Frederic Weisbecker <frederic@xxxxxxxxxx> wrote: > > Le Mon, Oct 02, 2023 at 07:51:10AM +0530, Neeraj upadhyay a écrit : > > > And if this works, can we then remove srcu_invoke_callbacks() self-requeue? > > > If queued several times before it actually fires, it will catch the latest > > > grace period's end. And if queued while the callback runs, it will re-run. > > > > > > > This makes sense, but not sure for non-wq context which link [1] mentions, > > whether it needs it. > > > > > Also why do we have sdp->srcu_invoke_callbacks ? Is that workqueue re-entrant? > > > > > I think you mean sdp->srcu_cblist_invoking ? > > > > There was a prior discussion on this [1], where Paul mentions about > > non-wq context. > > Actually I realize that if we remove that, then two srcu_invoke_callbacks() > can race in such a way that an srcu barrier callback may execute before the I wasn't aware that two srcu_invoke_callbacks() for same sdp can execute in parallel. My understanding was, any new callback will be handled by the same kworker, once it completes the current executing one [1]. Maybe I am missing your point here? [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/workqueue.c?h=v6.6-rc4#n1081 > callbacks it is supposed to follow. And if we must keep > sdp->srcu_cblist_invoking then we must also keep self-requeuing. > > I'll add a comment about that instead of removing. > > Thanks. > > > > > > > > > > Thanks > > Neeraj > > > > [1] https://lkml.org/lkml/2020/11/19/1065 > > > > > Thanks.
