Re: [PATCH v3] PCI: starfive: Enable PCIe controller's runtime PM before probing host bridge

[Bjorn Helgaas <helgaas@kernel.org>]

On Tue, Oct 15, 2024 at 10:09:01AM +0530, Manivannan Sadhasivam wrote:
> On Mon, Oct 14, 2024 at 01:08:41PM -0500, Bjorn Helgaas wrote:
> > On Mon, Oct 14, 2024 at 11:18:17PM +0530, Manivannan Sadhasivam wrote:
> > > On Mon, Oct 14, 2024 at 12:23:21PM -0500, Bjorn Helgaas wrote:
> > > > On Mon, Oct 14, 2024 at 09:26:07AM -0700, Mayank Rana wrote:
> > > > > PCIe controller device (i.e. PCIe starfive device) is parent to PCIe host
> > > > > bridge device. To enable runtime PM of PCIe host bridge device (child
> > > > > device), it is must to enable parent device's runtime PM to avoid seeing
> > > > > the below warning from PM core:
> > > > > 
> > > > > pcie-starfive 940000000.pcie: Enabling runtime PM for inactive device
> > > > > with active children
> > > > > 
> > > > > Fix this issue by enabling starfive pcie controller device's runtime PM
> > > > > before calling pci_host_probe() in plda_pcie_host_init().
> > > > > 
> > > > > Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
> > > > > Signed-off-by: Mayank Rana <quic_mrana@quicinc.com>
> > > > > Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
> > > > 
> > > > I want this in the same series as Krishna's patch to turn on runtime
> > > > PM of host bridges.  That's how I know they need to be applied in
> > > > order.  If they're not in the same series, they're likely to be
> > > > applied out of order.
> > > 
> > > There is no harm in applying this patch on its own. It fixes a legit
> > > issue of enabling the parent runtime PM before the child as required
> > > by the PM core. Rest of the controller drivers follow the same
> > > pattern.
> > > 
> > > I fail to understand why you want this to be combined with Krishna's
> > > patch. That patch is only a trigger, but even without that patch the
> > > issue still exists (not user visible ofc).
> > 
> > I don't want it *combined* with Krishna's patch.
> > 
> > I want this applied *before* Krishna's patch because if we apply
> > Krishna's patch first, we have some interval where we report the
> > "Enabling runtime PM for inactive device with active children" error.
> 
> No, I was asking why can't this be applied *first* and then
> Kirshna's patch? Why do you want this to be again resent in a
> separate series?

It can certainly be applied first.  It's just that when they're posted
separately, even with a comment in the email discussion to say "please
apply A before B", it's extra work for whoever merges them and it
makes it more likely to make mistakes like applying them in the wrong
order, or one without the other, or putting them on separate branches.

The usual way to say "B depends on A" is to post them together in a
series with A followed by B.

Bjorn