Re: [PATCH] i2c: i2c-ocores: Use devm_request_and_ioremap()

[Jean Delvare <khali-PUYAD+kWke1g9hUCZPvPmw@public.gmane.org>]

On Mon, 15 Oct 2012 09:21:41 +0200, Thierry Reding wrote:
> On Mon, Oct 15, 2012 at 09:10:15AM +0200, Jean Delvare wrote:
> > On Wed,  8 Aug 2012 08:54:32 +0200, Thierry Reding wrote:
> > > Replacing the devm_request_mem_region() and devm_ioremap_nocache() calls
> > > by a single call to devm_request_and_ioremap() simplifies the code.
> > > 
> > > Signed-off-by: Thierry Reding <thierry.reding-RM9K5IK7kjKj5M59NBduVrNAH6kLmebB@public.gmane.org>
> > > ---
> > >  drivers/i2c/busses/i2c-ocores.c | 15 +++------------
> > >  1 file changed, 3 insertions(+), 12 deletions(-)
> > > 
> > > diff --git a/drivers/i2c/busses/i2c-ocores.c b/drivers/i2c/busses/i2c-ocores.c
> > > index bffd550..1fad4ae 100644
> > > --- a/drivers/i2c/busses/i2c-ocores.c
> > > +++ b/drivers/i2c/busses/i2c-ocores.c
> > > @@ -283,18 +283,9 @@ static int __devinit ocores_i2c_probe(struct platform_device *pdev)
> > >  	if (!i2c)
> > >  		return -ENOMEM;
> > >  
> > > -	if (!devm_request_mem_region(&pdev->dev, res->start,
> > > -				     resource_size(res), pdev->name)) {
> > > -		dev_err(&pdev->dev, "Memory region busy\n");
> > > -		return -EBUSY;
> > > -	}
> > > -
> > > -	i2c->base = devm_ioremap_nocache(&pdev->dev, res->start,
> > > -					 resource_size(res));
> > > -	if (!i2c->base) {
> > > -		dev_err(&pdev->dev, "Unable to map registers\n");
> > > -		return -EIO;
> > > -	}
> > > +	i2c->base = devm_request_and_ioremap(&pdev->dev, res);
> > > +	if (!i2c->base)
> > > +		return -EADDRNOTAVAIL;
> > 
> > I would suggest sticking to -EBUSY, AFAIK EADDRNOTAVAIL is for network
> > address errors.
> 
> Actually the kerneldoc comment for devm_request_and_ioremap()
> (lib/devres.c) explicitly lists this as the error code to return on
> failure. EBUSY could be misleading since the error could just as well
> come from a failed ioremap().

Ah, my bad, I missed that comment.

It's probably not relevant as I'm not in charge of the devm
infrastructure, but I don't like it. There are 3 different possible
causes of failure for devm_request_and_ioremap, and the caller can't
tell which one happened. We have ERR_PTR for cases like this... OTOH I
guess the caller wouldn't behave differently based on the error code,
so it probably doesn't matter that much. So...

Acked-by: Jean Delvare <khali-PUYAD+kWke1g9hUCZPvPmw@public.gmane.org>

Wolfram, can you please pick this patch?

Thanks,
-- 
Jean Delvare