Re: [PATCH 00/14] SCI clocks cleanup

[Simon Horman <horms@verge.net.au>]

On Mon, Sep 14, 2015 at 03:14:22PM +0300, Laurent Pinchart wrote:
> Hello,
> 
> The SCI driver currently handles two clocks, an interface clock named sci_ick
> and a functional clock named sci_fck. Studying the datasheets of the SH and
> ARM SoCs that incorportate (H)SCI(F)([AB]) instances showed (un)surprisingly
> that the hardware doesn't have a separate controllable interface clock.
> 
> All the platforms that declare an interface clock for the SCI set it to the
> clock used as the SCI functional clock. The two clocks can thus be merged on
> the driver side, which is what this patch series does. The resulting clock is
> called "fck", and all SH and ARM users (both DT and non-DT) are fixed to name
> their SCI clocks appropriately.
> 
> Support for the "sci_ick" name is kept in the sh-sci driver to ensure DT
> backward compatibility, and support for the "peripheral_clk" clock to not
> break SH platforms that don't declare device-specific SCI clocks. The later
> can be removed when all SH platforms will declare their SCI clocks properly.
> 
> The patches have been developed for an ancien (v3.x for those who were born)
> kernel and rebased on top of Simon's master branch. I've only compile-tested
> them after the rebase. Geert, I believe this series is a good preliminary
> cleanup for the SCI baud rate generator clock support. Could you give it a try
> as part of your work on that ?

FWIW, given the explanation above, particularly that regarding backwards
compatibility for DT, I am comfortable with these changes.