On Thu, Feb 21, 2019 at 1:56 PM Andy Shevchenko <[email protected]> wrote: > > On Thu, Feb 21, 2019 at 6:27 PM Bartosz Golaszewski <[email protected]> wrote: > > > > From: Bartosz Golaszewski <[email protected]> > > > > We only build devm_ioremap_resource() if HAS_IOMEM is selected, so this > > dependency must cascade down to devm_platform_ioremap_resource(). > > > +#ifdef CONFIG_HAS_IOMEM > > void __iomem *devm_platform_ioremap_resource(struct platform_device *pdev, > > unsigned int index) > > { > > @@ -96,6 +97,7 @@ void __iomem *devm_platform_ioremap_resource(struct > > platform_device *pdev, > > return devm_ioremap_resource(&pdev->dev, res); > > } > > EXPORT_SYMBOL_GPL(devm_platform_ioremap_resource); > > +#endif /* CONFIG_HAS_IOMEM */ > > What about declaration in *.h? > > Perhaps you may just do this inside the function?
Just a drive by comment, but IMO we should just get rid of HAS_IOMEM altogether. It is really just a !UML option as I think every other arch enables it. If folks really want to disable drivers on UML, just disable the subsystems. Though now with kunit mocking, there's a desire to enable HAS_IOMEM on UML, too. Rob
