On Wed, Dec 21, 2011 at 08:55:09PM +0100, Janusz Krzysztofik wrote:

On Wednesday 21 of December 2011 at 20:09:45, Tony Lindgren wrote:

quoted

* Janusz Krzysztofik [off-list ref] [111220 13:39]:

quoted

Don't use Amstrad Delta custom I/O functions once GPIO interface is
available for the underlying hardware.

While requesting and initializing GPIO pins used, also take care of one
extra pin KEYBRD_DATAOUT which, even if not used by the driver, belongs
to the device and affects its functioning.

Once done, move the driver initialization back to the device_initcall
level, reverting the temporary chane introduced with patch 1/7 "ARM:
OMAP1: ams-delta: register latch dependent devices later".  That change
is no longer required once the driver takes care of registering used
GPIO pins, and it's better to initialize the device before others using
the latch2 based GPIO pins, otherwise a garbage is reported on boot,
perhaps due to random data already captured by the FIQ handler while the
keyboard related latch bits are written with random values during
initialization of those other latch2 dependent devices.

Depends on patch 2/7 "ARM: OMAP1: ams-delta: convert latches to
basic_mmio_gpio"

Signed-off-by: Janusz Krzysztofik <redacted>

I'm assuming Dmitry's ack for an earlier version of this patch also
covers this one when applying.

Hi Dmitry,
Can we assume your ack still valid for this version?

Yes, please merge with your other changes.

BTW, if you could rename :static struct gpio _gpios" to "ams_delta_gpios"
that would be even better.

Thanks!

-- 
Dmitry