Hello Alexandre,

On Wed, Aug 12, 2020 at 10:47:28AM +0200, Alexandre Belloni wrote:
> On 12/08/2020 10:32:04+0200, Uwe Kleine-König wrote:
> > On Wed, Aug 12, 2020 at 09:20:02AM +0100, Lee Jones wrote:
> > > On Wed, 12 Aug 2020, Uwe Kleine-König wrote:
> > > >  	atmel_pwm->clk = devm_clk_get(&pdev->dev, NULL);
> > > >  	if (IS_ERR(atmel_pwm->clk))
> > > > -		return PTR_ERR(atmel_pwm->clk);
> > > > +		return dev_err_probe(&pdev->dev, PTR_ERR(atmel_pwm->clk),
> > > > +				     "Failed to get clock\n");
> > > 
> > > Isn't dev_err_probe() only useful for drivers handling -EPROBE_DEFER?
> > 
> > devm_clk_get() might return -EPROBE_DEFER.
> 
> If it did, you wouldn't be able to print this message.

Why that? It probably won't make it to the console immediately, but once
the clk is available the log buffer should be pushed out, shouldn't it?

> I' not sure it is worth adding so many checks for errors that will
> never happen.

I'm sure this train of thought is unsustainable. And people will copy
this code to platforms where this assumption might even be more wrong
than on at91. 

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |