Re: [PATCH v5 05/11] PCI: cadence: Extract link setup sequence from cdns_pcie_host_setup()

[Dan Carpenter <dan.carpenter@xxxxxxxxxx>]

On Tue, May 14, 2024 at 03:15:34PM +0200, Thomas Richard wrote:
> On 4/16/24 16:16, Dan Carpenter wrote:
> > On Tue, Apr 16, 2024 at 03:29:54PM +0200, Thomas Richard wrote:
> >> diff --git a/drivers/pci/controller/cadence/pcie-cadence-host.c b/drivers/pci/controller/cadence/pcie-cadence-host.c
> >> index 5b14f7ee3c79..93d9922730af 100644
> >> --- a/drivers/pci/controller/cadence/pcie-cadence-host.c
> >> +++ b/drivers/pci/controller/cadence/pcie-cadence-host.c
> >> @@ -497,6 +497,30 @@ static int cdns_pcie_host_init(struct device *dev,
> >>  	return cdns_pcie_host_init_address_translation(rc);
> >>  }
> >>  
> >> +int cdns_pcie_host_link_setup(struct cdns_pcie_rc *rc)
> >> +{
> >> +	struct cdns_pcie *pcie = &rc->pcie;
> >> +	struct device *dev = rc->pcie.dev;
> >> +	int ret;
> >> +
> >> +	if (rc->quirk_detect_quiet_flag)
> >> +		cdns_pcie_detect_quiet_min_delay_set(&rc->pcie);
> >> +
> >> +	cdns_pcie_host_enable_ptm_response(pcie);
> >> +
> >> +	ret = cdns_pcie_start_link(pcie);
> >> +	if (ret) {
> >> +		dev_err(dev, "Failed to start link\n");
> >> +		return ret;
> >> +	}
> >> +
> >> +	ret = cdns_pcie_host_start_link(rc);
> >> +	if (ret)
> >> +		dev_dbg(dev, "PCIe link never came up\n");
> > 
> > If we're going to ignore this error the message should be a dev_err()
> > at least.
> 
> Hello Dan,
> 
> In fact it could not be really an error.
> If you physically don't have a device on the PCIe bus,
> cdns_pcie_host_start_link() will not return 0.
> 
> So if we use dev_err(), we will always have the error if there is no
> device on the PCIe bus.

Ah okay.  Thanks for looking at this.  It feels like maybe
cdns_pcie_host_start_link() should just check for that at the start and
return 0 instead of doing waiting 1 second and returning -ETIMEOUT.
But I don't know this code well enough to say if that's even possible.

regards,
dan carpenter