gpu: host1x: Fix bitshift/mask multipliers

gpu: host1x: Fix bitshift/mask multipliers

Some parts of Host1x uses BIT_WORD/BIT_MASK/BITS_PER_LONG to calculate
register or field offsets. This worked fine on ARMv7, but now that
BITS_PER_LONG is 64 but our registers are still 32-bit things are
broken.

Fix by replacing..
- BIT_WORD with (x / 32)
- BIT_MASK with BIT(x % 32)
- BITS_PER_LONG with 32

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>