[PATCH 00/10] clk: mvebu: restructure SoC clock drivers

[thomas.petazzoni@free-electrons.com (Thomas Petazzoni)]

Sebastian, Jason, Andrew, Gregory,

On Sat, 11 May 2013 03:08:00 +0200, Sebastian Hesselbarth wrote:
> Marvell EBU SoCs share some common clock functions that register core
> clocks or clock gates from SoC specific data. These clock drivers are
> organized by function and over time the drivers filled up with #ifdefs
> to separate different SoC specific code.
> 
> This patch set first adds a new element to clock gate descriptors to
> allow to pass clk flags per clock gate. (Patch 1)
> 
> Then it restructures MVEBU clock drivers into common clock setup and
> SoC specific files and Kconfig options. The driver's init functions
> are also registered to get called with of_clk_init. (Patches 2, 3-6)
> 
> It then switches Dove, Kirkwood, and Armada 370/XP to make use of
> of_clk_init instead of mvebu_clocks_init. (Patches 7-9)
> 
> Finally, all obsolete files and Kconfig options are removed. (Patch 10)
> 
> It has been tested on CuBox (Dove) and Dockstar (Kirkwood) by me.
> Armada 370 and XP are compile tested only and should get their Tested-by
> from Thomas and Gregory, respectively.
> 
> @Mike: As this only touches MVEBU related code, I suggest to take it
> all through the corresponding ARM branch.
> 
> Sebastian Hesselbarth (10):
>   clk: mvebu: introduce per-clock-gate flags
>   clk: mvebu: add common clock functions for core clk and clk gating
>   clk: mvebu: add Dove SoC-centric clock init
>   clk: mvebu: add Kirkwood SoC-centric clock init
>   clk: mvebu: add Armada 370 SoC-centric clock init
>   clk: mvebu: add Armada XP SoC-centric clock init
>   ARM: dove: move DT boards to SoC-centric clock init
>   ARM: kirkwood: move DT boards to SoC-centric clock init
>   ARM: mvebu: move DT boards to SoC-centric clock init
>   clk: mvebu: desintegrate obsolete file

Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

I've tested on Armada 370 DB (after reverting the ARM core patch that
prevents this platform from booting) and the Armada XP DB. Both boards
boot fine, and detect PCIe devices as they should.

There is some conflict between this set of commits and the two commits
that modify the PCIe clocks in my PCIe series. I've pushed a branch at
https://github.com/MISL-EBU-System-SW/mainline-public/tree/marvell-pcie-v10-and-seb-clks
which has the mvebu PCIe stuff + Sebastian patches on top of it, with
the conflicts resolved. The tricky thing is that since Sebastian is
adding new clock files, and then removing the old ones, it's pretty
easy to "loose" the changes the two PCIe clock patches are doing. The
branch above works on Armada 370/XP, so you can compare the clock files
if needed.

Of course, I'm willing to do another final test once you've done the
merged to confirm that everything is alright.

I haven't done a detailed review of the patches, but I definitely agree
with the principle, and I trust Sebastian for having done the right
thing in here.

Thanks!

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com