On Wed, Jul 6, 2022 at 8:01 AM Marc Zyngier <maz@xxxxxxxxxx> wrote: > > On Tue, 05 Jul 2022 05:53:03 +0100, > "Lad, Prabhakar" <prabhakar.csengg@xxxxxxxxx> wrote: > > > > On Mon, Jul 4, 2022 at 5:16 PM Andy Shevchenko > > <andy.shevchenko@xxxxxxxxx> wrote: > > > > > > On Sun, Jul 3, 2022 at 9:43 PM Lad Prabhakar <prabhakar.csengg@xxxxxxxxx> wrote: > > > > > > > > Allow free() callback to be overridden from irq_domain_ops for > > > > hierarchical chips. > > > > > > > > This allows drivers to free up resources which are allocated during > > > > child_to_parent_hwirq()/populate_parent_alloc_arg() callbacks. > > > > > > > > On Renesas RZ/G2L platform a bitmap is maintained for TINT slots, a slot > > > > is allocated in child_to_parent_hwirq() callback which is freed up in free > > > > callback hence this override. > > > > > > Hmm... To me this sounds asymmetrical. We alloc something in another > > > callback, which is not what this free is for. Perhaps it should be an > > > optional > > > > > > free_populated_parent_arg() or alike? > > > > > @Marc your thoughts? > > I think there are enough optional callbacks, and we don't currently > have a clear picture of how this may be used in the future, specially > based on a sample of *one*. > > Let's get it in as is, and tweak things as we go, should another user > require a slightly different behaviour. This also saves us the debate > around the naming, which is always pretty useless. > Thanks, I will repost it as is. Cheers, Prabhakar
