Re: [PATCH v3] PCI: pci-epf-test: Register notifier if only core_init_notifier is enabled

[Bjorn Helgaas <helgaas@kernel.org>]

On Tue, Nov 01, 2022 at 06:31:07PM +0900, Kunihiko Hayashi wrote:
> Hi Bjorn,
> 
> On 2022/10/29 2:06, Bjorn Helgaas wrote:
> > On Thu, Aug 25, 2022 at 06:01:01PM +0900, Kunihiko Hayashi wrote:
> > > Need to register pci_epf_test_notifier function event if only
> > > core_init_notifier is enabled.
> > > 
> > > Fixes: 5e50ee27d4a5 ("PCI: pci-epf-test: Add support to defer core
> > initialization")
> > > Signed-off-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
> > > Acked-by: Om Prakash Singh <omp@nvidia.com>
> > > Acked-by: Kishon Vijay Abraham I <kishon@ti.com>
> > > ---
> > >   drivers/pci/endpoint/functions/pci-epf-test.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > This patch is a part of series "PCI: endpoint: Fix core_init_notifier
> > feature".
> > > The rest of the patches have been withdrawn.
> > > 
> > > Changes since v2:
> > > - Add Acked-by lines
> > > 
> > > Changes since v1:
> > > - Add Acked-by lines
> > > 
> > > diff --git a/drivers/pci/endpoint/functions/pci-epf-test.c
> > b/drivers/pci/endpoint/functions/pci-epf-test.c
> > > index 36b1801a061b..55283d2379a6 100644
> > > --- a/drivers/pci/endpoint/functions/pci-epf-test.c
> > > +++ b/drivers/pci/endpoint/functions/pci-epf-test.c
> > > @@ -979,7 +979,7 @@ static int pci_epf_test_bind(struct pci_epf *epf)
> > >   	if (ret)
> > >   		epf_test->dma_supported = false;
> > > -	if (linkup_notifier) {
> > > +	if (linkup_notifier || core_init_notifier) {
> > >   		epf->nb.notifier_call = pci_epf_test_notifier;
> > >   		pci_epc_register_notifier(epc, &epf->nb);
> > 
> > Why does pci_epc_register_notifier() even exist?  It's not used at all
> > except for this test code.
> > 
> > It would be better if infrastructure like this were connected with
> > some user of it.
>
> This call was added by the commit 5779dd0a7dbd
> ("PCI: endpoint: Use notification chain mechanism to notify EPC events to EPF").
> 
> I haven't followed the discussion, however, this commit say: "This
> will also enable to add more events (in addition to linkup) in the
> future."

5779dd0a7dbd was 2.5 years ago.  It was probably a mistake to merge it
then.  Usually we add infrastructure when we have a user for it,
either in the very same patch or at least in a subsequent patch of the
series.  

But Lorenzo has already merged this, so I guess this is moot.

He probably wouldn't object to a second patch that removes the fixed
infrastructure if nobody uses it.  We can easily resurrect it if a
need arises, and if that happens, we'll be glad that it has been
fixed!

Bjorn