Re: [PATCH v2 3/4] PCI: imx6: Convert to agnostic GPIO API

[Frank Li <Frank.li@nxp.com>]

On Tue, Apr 23, 2024 at 04:16:36PM -0400, Frank Li wrote:
> On Tue, Apr 23, 2024 at 11:03:59PM +0300, Andy Shevchenko wrote:
> > On Tue, Apr 23, 2024 at 03:56:56PM -0400, Frank Li wrote:
> > > On Tue, Apr 23, 2024 at 08:19:06PM +0300, Andy Shevchenko wrote:
> > 
> > ...
> > 
> > > > +	imx6_pcie->reset_gpiod =
> > > > +		devm_gpiod_get_optional(dev, "reset",
> > > > +			imx6_pcie->gpio_active_high ? GPIOD_OUT_HIGH : GPIOD_OUT_LOW);
> > > > +	if (IS_ERR(imx6_pcie->reset_gpiod))
> > > > +		return dev_err_probe(dev, PTR_ERR(imx6_pcie->reset_gpiod),
> > > > +				     "unable to get reset gpio\n");
> > > 
> > > Small problem here. err message "unable to get reset gpio\n" will print
> > > when -EPROBE_DEFER happen. EPROBE_DEFER is quite common when use i2c
> > > expand gpio chip.
> > 
> > I'm not sure how you come to this conclusion. Can you elaborate, please?
> > P.S> I do not see a problem as described.
> 
> If i2c gpio-expander driver have not load when imx6_pcie probe, I supposed
> devm_gpiod_get_optional() will return -EPROBE_DEFER, 
> 
> if (IS_ERR(imx6_pcie->reset_gpiod)) should be true. then dev_err_probe()
> will run and print "unable to get reset gpio\n" with error code
> -EPROBE_DEFER.

Sorry for that. dev_err_probe() already consider this. Please forget my
comments.

dev_err_probe() 
{
	if (err != -EPROBE_DEFER) {
		dev_err(dev, "error %pe: %pV", ERR_PTR(err), &vaf);
	}
}

So:

Reviewed-by: Frank Li <Frank.Li@nxp.com>

> 
> driver framework will retry imx6_pcie probe again when a new device appear.
> it may retry sevial times utill i2c gpio-expander driver probe success or
> timeout.
> 
> Frank
> 
> > 
> > -- 
> > With Best Regards,
> > Andy Shevchenko
> > 
> >