On 06/03/11 10:42, Grant Likely wrote:
On Fri, Jun 3, 2011 at 11:18 AM, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Fri, Jun 03, 2011 at 11:04:52AM -0600, Grant Likely wrote:
I ended up not pushing this one to Linus. Turns out it causes other breakage on other platforms that don't include include/linux/gpio.h. Since I don't have confidence that I'll be able to find all the offenders, I'm dropping it. I recommend making any drivers that are
So, this originally came about because I pushed back on adding random dependencies like this for features which are pretty much optional in drivers - their use of GPIOs is totally optional and the dependencies are just too fragile, leading to noise with all the randconfigs. It seems better to get the architectures to keep up with enhancements to gpiolib (or convert to it) than to have to worry about this in drivers.
Fair enough. Randy, if you or someone else can check that all GPIOF_ users have the required #include <linux/gpio.h>, then I'm okay with this patch.
OK, I'll look at that.
Do you have any examples of builds that failed with this patch?
thanks,