By not definining a custom gpio_to_irq, the default gpiolib version is
used, allowing platform consolidation.

irq_to_gpio is deprecated and in the process of being removed. Make that
happen now for ARM Tegra.

This also partially fixes the Tegra build; it was broken because gpio.h
referred to EINVAL, which wasn't always defined when <mach/gpio.h> was
included.
Signed-off-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>