Re: [RFC 04/42] drivers/dma: don't check resource with devm_ioremap_resource

[Wolfram Sang <wsa-z923LK4zBo2bacvFa/9K2g@public.gmane.org>]

On Fri, May 10, 2013 at 10:35:32AM -0600, Stephen Warren wrote:
> On 05/10/2013 02:16 AM, Wolfram Sang wrote:
> > devm_ioremap_resource does sanity checks on the given resource. No need to
> > duplicate this in the driver.
> 
> > diff --git a/drivers/dma/tegra20-apb-dma.c b/drivers/dma/tegra20-apb-dma.c
> 
> >  	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > -	if (!res) {
> > -		dev_err(&pdev->dev, "No mem resource for DMA\n");
> > -		return -EINVAL;
> > -	}
> > -
> >  	tdma->base_addr = devm_ioremap_resource(&pdev->dev, res);
> 
> One issue here is that it's not obvious just from reading the code
> that's left behind that the "missing" error-checking of the
> platform_get_resource() return value is OK because
> devm_ioremap_resource() will check it "for us". Everyone now has to
> mentally maintain a list of exceptions where it's OK not to error-check.

My goal is to make not-checking the standard case with devm.

> A similar situation exists for e.g. kzalloc already spewing an error
> when allocations fail, and so drivers don't need to print a diagnostic
> in that case, but it's helpful if they do in most other cases, but
> that's another issue.

My goal is to make drivers-not-printing-errors the standard case with devm :)

> Would it be better to introduce a new devm_ioremap_pdev_resource(pdev,
> index) that replaced both those two API calls with a single one. That
> way, only the author/reader of the new devm_ioremap_pdev_resource()
> would have to remember that caveat, and a single comment could be added
> so people unfamiliar with the code remember, without duplicating the
> comment in every driver.

I see two steps here:

1) make devm interface consistent
2) ease the devm interface with platform drivers

I am currently working on the first step.

Thanks,

   Wolfram