On Wed, Jul 06, 2011 at 08:36:48AM +0200, Sascha Hauer wrote:

On Tue, Jul 05, 2011 at 11:27:55AM -0600, Grant Likely wrote:

quoted

On Tue, Jul 5, 2011 at 9:05 AM, Shawn Guo [off-list ref] wrote:

quoted

The patch removes all the uses of cpu_is_mx1(). ?Instead, it uses
the .id_table of platform_driver to distinguish the uart device type.

A couple of !cpu_is_mx1() logic gets turned into IS_IMX_UART, as the
codes wrapped there are really IMX specific.

It also removes macro MX1_UCR3_REF25 and MX1_UCR3_REF30 which are
not used anywhere.

Signed-off-by: Shawn Guo <redacted>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Alan Cox <redacted>

Acked-by: Grant Likely <redacted>

Who's tree is this one going to get merged through?

Before we merge this through any tree I'd like to have a policy how
the bindings are named. With the gpio part we agreed that we name
the bindings after the oldest i.MX they are compatible with. For example
the last change in the gpios was on the i.MX31, so every newer i.MX
claims to be compatible with this the i.MX31.

The patch below instead follows the approach that the currently newest
i.MX gets a generic name (imx-uart instead of imx21-uart). This decision
may look silly when another change to the uart core comes and we have
given the generic name to some arbitrary version of the core.

I really like to have a consistent scheme across all devices and we have
the same problem with spi, nand and several others.

Makes sense.  Will fix it in v3.

-- 
Regards,
Shawn