Messages in this thread | | | From | Tomasz Figa <> | Subject | Re: [PATCH 0/2] ARM: Exynos4: Migrate to common clock framework. | Date | Wed, 03 Oct 2012 09:26:50 +0200 |
| |
Hi Chander, Thomas,
On Monday 01 of October 2012 17:39:19 chander.kashyap@linaro.org wrote: > From: Thomas Abraham <thomas.abraham@linaro.org> > > This patch series migrates Exynos4 clock support to common clock > framework. The first patch in this series removes the existing Exynos4 > clock support that uses the Samsung specific clock framework. The second > patch in this series add Exynos4 clock support using common clock > framework. > > Thomas Abraham (2): > ARM: Exynos4: Remove Samsung clock type support > ARM: Exynos4: Register clocks via common clock framework
I think the order of changes is a little bit off here: - patch 1 will break all exynos4-based boards (what about bisects?) - patch 2 will be still broken until all related drivers get converted to use clk_prepare(_enable) and clk_(disable_)unprepare.
Shouldn't the order be exactly opposite, i.e.: - all the patches for prepare/unprepare first - then the patch adding common clock frameworks support for exynos4 (disabling the old clock code) - and finally the patch removing remaining (disabled by previous patch) code.
Also, I assume that these patches doesn't consider native device tree support (without auxdata, using OF-based clock lookup), correct me if I'm wrong. If I'm right, since Exynos SoCs are going to be DT-only, is there really a point for adding common clock framework support for non-DT platforms (which are going to be eventually dropped anyway)?
Best regards, -- Tomasz Figa Samsung Poland R&D Center
| |