If the link is not up till the pwrctl drivers enable power to endpoints
then cur_bus_speed will not be updated with correct speed.

As part of rescan, pci_bus_add_devices() will be called and as part of
it update the link bus speed.

Signed-off-by: Krishna Chaitanya Chundru <krishna.chundru@oss.qualcomm.com>
---
 drivers/pci/bus.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/pci/bus.c b/drivers/pci/bus.c
index 98910bc0fcc4..994879071d4c 100644
--- a/drivers/pci/bus.c
+++ b/drivers/pci/bus.c
@@ -432,6 +432,9 @@ void pci_bus_add_devices(const struct pci_bus *bus)
 	struct pci_dev *dev;
 	struct pci_bus *child;
 
+	if (bus->self)
+		pcie_update_link_speed((struct pci_bus *)bus);
+
 	list_for_each_entry(dev, &bus->devices, bus_list) {
 		/* Skip already-added devices */
 		if (pci_dev_is_added(dev))

-- 
2.34.1

[+cc Bartosz since this is related to pwrctrl]

On Thu, Mar 13, 2025 at 05:10:08PM +0530, Krishna Chaitanya Chundru wrote:
> If the link is not up till the pwrctl drivers enable power to endpoints
> then cur_bus_speed will not be updated with correct speed.
> 
> As part of rescan, pci_bus_add_devices() will be called and as part of
> it update the link bus speed.
> 
> Signed-off-by: Krishna Chaitanya Chundru <krishna.chundru@oss.qualcomm.com>
> ---
>  drivers/pci/bus.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/pci/bus.c b/drivers/pci/bus.c
> index 98910bc0fcc4..994879071d4c 100644
> --- a/drivers/pci/bus.c
> +++ b/drivers/pci/bus.c
> @@ -432,6 +432,9 @@ void pci_bus_add_devices(const struct pci_bus *bus)
>  	struct pci_dev *dev;
>  	struct pci_bus *child;
>  
> +	if (bus->self)
> +		pcie_update_link_speed((struct pci_bus *)bus);

Why the cast?  I guess to discard "const"?  I would prefer to avoid
that if we can.

The only existing callers of pcie_update_link_speed() are from bwctrl,
where it obviously makes sense because bwctrl manages link speed, and
from pci_set_bus_speed() in the pci_alloc_child_bus() path.

If we call pcie_update_link_speed() here, do we still need the
pci_set_bus_speed() in pci_alloc_child_bus()?

Bartosz, so you have any suggestions?  Is there a point in pwrctrl
where it would make sense to do this, similar to what bwctrl does?

>  	list_for_each_entry(dev, &bus->devices, bus_list) {
>  		/* Skip already-added devices */
>  		if (pci_dev_is_added(dev))
> 
> -- 
> 2.34.1
>

On Thu, Mar 13, 2025 at 12:00:01PM -0500, Bjorn Helgaas wrote:
> [+cc Bartosz since this is related to pwrctrl]
> 
> On Thu, Mar 13, 2025 at 05:10:08PM +0530, Krishna Chaitanya Chundru wrote:
> > If the link is not up till the pwrctl drivers enable power to endpoints
> > then cur_bus_speed will not be updated with correct speed.
> > 
> > As part of rescan, pci_bus_add_devices() will be called and as part of
> > it update the link bus speed.
> > 
> > Signed-off-by: Krishna Chaitanya Chundru <krishna.chundru@oss.qualcomm.com>
> > ---
> >  drivers/pci/bus.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/drivers/pci/bus.c b/drivers/pci/bus.c
> > index 98910bc0fcc4..994879071d4c 100644
> > --- a/drivers/pci/bus.c
> > +++ b/drivers/pci/bus.c
> > @@ -432,6 +432,9 @@ void pci_bus_add_devices(const struct pci_bus *bus)
> >  	struct pci_dev *dev;
> >  	struct pci_bus *child;
> >  
> > +	if (bus->self)
> > +		pcie_update_link_speed((struct pci_bus *)bus);
> 
> Why the cast?  I guess to discard "const"?  I would prefer to avoid
> that if we can.
> 
> The only existing callers of pcie_update_link_speed() are from bwctrl,
> where it obviously makes sense because bwctrl manages link speed, and
> from pci_set_bus_speed() in the pci_alloc_child_bus() path.
> 
> If we call pcie_update_link_speed() here, do we still need the
> pci_set_bus_speed() in pci_alloc_child_bus()?
> 
> Bartosz, so you have any suggestions?  Is there a point in pwrctrl
> where it would make sense to do this, similar to what bwctrl does?
> 

I'm not Bartosz, but I could speak for pwrctrl.

This should really be part of pci_pwrctrl_notify() in pwrctrl driver. Pwrctrl
driver is already getting notified with device additions and removals. So the
update can be performed once the device addition happens.

- Mani

-- 
மணிவண்ணன் சதாசிவம்

s/update/Update/ in subject to match others