Re: [RFC 2/6] ipv4: add lockdep condition to fix for_each_entry

[Joel Fernandes]

On Mon, Jun 3, 2019 at 2:42 AM Pavel Machek <pavel@xxxxxxx> wrote:
>
> On Sun 2019-06-02 08:24:35, Joel Fernandes wrote:
> > On Sun, Jun 2, 2019 at 8:20 AM Joel Fernandes <joel@xxxxxxxxxxxxxxxxx> wrote:
> > >
> > > On Sun, Jun 2, 2019 at 3:00 AM Pavel Machek <pavel@xxxxxxx> wrote:
> > > >
> > > > On Sat 2019-06-01 18:27:34, Joel Fernandes (Google) wrote:
> > > > > Signed-off-by: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>
> > > >
> > > > This really needs to be merged to previous patch, you can't break
> > > > compilation in middle of series...
> > > >
> > > > Or probably you need hlist_for_each_entry_rcu_lockdep() macro with
> > > > additional argument, and switch users to it.
> > >
> > > Good point. I can also just add a temporary transition macro, and then
> > > remove it in the last patch. That way no new macro is needed.
> >
> > Actually, no. There is no compilation break so I did not follow what
> > you mean. The fourth argument to the hlist_for_each_entry_rcu is
> > optional. The only thing that happens is new lockdep warnings will
> > arise which later parts of the series fix by passing in that fourth
> > argument.
>
> Sorry, I missed that subtlety. Might be worth it enabling the lockdep
> warning last in the series...

Good idea, will do! Thanks.