[U-Boot] [PATCH 2/2] Add support the Avionic Design Meerkat COM and Kein Baseboard

[Thierry Reding]

On Mon, Feb 29, 2016 at 09:48:21AM -0700, Stephen Warren wrote:
> On 02/29/2016 03:20 AM, Julian Scheel wrote:
> >Hi Stephen,
> >
> >On 23.02.2016 19:03, Stephen Warren wrote:
> >>On 02/23/2016 05:59 AM, Julian Scheel wrote:
> >>>Add support for platforms based on the Meerkat COM module. Includes
> >>>support
> >>>for the minimal reference platform called Kein Baseboard, which in
> >>>fact is
> >>>sufficient to run most existing Meerkat carriers.
> >>
> >>>diff --git a/arch/arm/dts/tegra124-meerkat.dtsi
> >>>b/arch/arm/dts/tegra124-meerkat.dtsi
> >>
> >>>+    aliases {
> >>>+        i2c0 = "/i2c at 7000c000";
> >>>+        i2c1 = "/i2c at 7000c400";
> >>>+        i2c2 = "/i2c at 7000c500";
> >>>+        i2c3 = "/i2c at 7000c700";
> >>>+        i2c4 = "/i2c at 7000d000";
> >>>+        i2c5 = "/i2c at 7000d100";
> >>
> >>/i2c at 7000d000 is alias i2c0 on all other Tegra boards that have I2C
> >>aliases.
> >
> >we just had a quick discussion about this and actually we explicitly
> >changed it in our tree to a linear mapping of the controller addresses
> >to be in sync with what we have in the l4t kernel.
> >If it won't hurt you too much we'd like to keep it that way. Though if
> >you feel really bad about it, we can change it to use the same mapping
> >as Jetson uses.
> 
> That's probably fine; hopefully it won't be too confusing since aliases are
> supposed to be board-specific.
> 
> Let's make sure that the Linux kernel patches for this board have matching
> aliases and get Ack'd soon to make sure they don't diverge.

I don't mind having the mapping like this for Meerkat. For anything in
kernel space this should be completely irrelevant since the association
is done via parent-child relationships in DT anyway.

The only place where the numbering matters is if you want to use the I2C
busses from userspace (using i2c-dev) and in that case you're probably
better off if you use some sort of lookup mechanism to find the correct
I2C bus based on the MMIO address, instead of relying on hardcoded bus
numbers.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20160302/00f3a3fd/attachment.sig>