first bunch of rockchip clock changes for 4.6

Hi Mike, Stephen,

no new clock trees this time, but more clock ids and fixes for issues present for a longer time (not introduced during the 4.5-rc1 merge wind= ow)
that Stephen thought should go through the regular process.

Most "interesting" change might be that we now handle factors correctly=

according to their position in the clock-tree instead of registering th= em
individually during init - which became a clutch.

So if nothing stands out, please pull

Thanks
Heiko

The following changes since commit 92e963f50fc74041b5e9e744c330dca48e04= f08d:

Linux 4.5-rc1 (2016-01-24 13:06:47 -0800)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip.gi=
t tags/v4.6-rockchip-clk1

for you to fetch changes up to 36714529f8bbd4f8eaf93b50f4a64c52a24879aa= :

clk: rockchip: convert manually created factor clocks to the new type=
(2016-02-04 15:54:20 +0100)

----------------------------------------------------------------
Introduction of a factor type and a variant containing a gate to be able to also declare factor clocks in their correct place in the clock tree instead of having to register factor clocks in the init callback separately. And as always some more clock-ids and non-regression fixes for mistakes introduced in past kernel releases.

----------------------------------------------------------------
Alexander Kochetkov (2):
clk: rockchip: add hclk_cpubus to the list of rk3188 critical clo=
cks
clk: rockchip: Allow sclk_i2s0 and i2s0_frac to change their pare=
nts rate on rk3188

Heiko Stuebner (9):
clk: rockchip: rk3368: fix cpuclk mux bit of big cpu-cluster
clk: rockchip: rk3368: fix cpuclk core dividers
clk: rockchip: rk3368: fix parents of video encoder/decoder
clk: rockchip: rk3368: fix hdmi_cec gate-register
clk: rockchip: fix usbphy-related clocks
Merge branch 'v4.6-shared/clkids' into v4.6-clk/next
clk: rockchip: fix parent of hclk_vcodec on rk3036
clk: rockchip: add a factor clock type
clk: rockchip: convert manually created factor clocks to the new =
type

Pawe=C5=82 Jarosz (2):
clk: rockchip: Add new id for rk3066 tsadc clock
clk: rockchip: add tsadc clock on rk3066

Shawn Lin (2):
clk: rockchip: fix wrong mmc phase shift for rk3228
clk: rockchip: free memory in error cases when registering clock =
branches

Yakir Yang (1):
clk: rockchip: rk3036: enable the CLK_IGNORE_UNUSED flag for hclk=
_vio_bus

zhangqing (3):
clk: rockchip: rk3368: fix edp_24m parent
clk: rockchip: rk3368: enable the CLK_SET_RATE_PARENT flag for sp=
dif_8ch
clk: rockchip: rk3368: enable the CLK_SET_RATE_PARENT flag for i2=
s_2ch

 arch/arm/boot/dts/rk3288-veyron.dtsi          |  2 +-
 drivers/clk/rockchip/clk-rk3036.c             | 33 +++--------
 drivers/clk/rockchip/clk-rk3188.c             | 27 +++------
 drivers/clk/rockchip/clk-rk3228.c             | 38 +++----------
 drivers/clk/rockchip/clk-rk3288.c             | 38 ++++---------
 drivers/clk/rockchip/clk-rk3368.c             | 82 +++++++++++--------=
--------
 drivers/clk/rockchip/clk.c                    | 64 +++++++++++++++++++=
+-
 drivers/clk/rockchip/clk.h                    | 28 +++++++++
 include/dt-bindings/clock/rk3188-cru-common.h |  1 +
 9 files changed, 159 insertions(+), 154 deletions(-)

Hello Heiko,

Quoting Heiko Stuebner (2016-02-11 05:00:02)

io_bus

Thanks for sending the very readable PR. One question below,

We'll likely have critical clock and handoff clock support merged in the next few days. I'm starting to audit any new CLK_IGNORE_UNUSED flag usage to see if it should use either of those two flags. Is the above patch a candidate?

Thanks,
Mike

Hi Mike,

Am Donnerstag, 11. Februar 2016, 18:19:36 schrieb Michael Turquette:

the above clock is part of the interconnect. We don't model that yet, but I guess that would make it more a candidate for the handoff-type instead.

I'd guess for the Rockchip side, everything that is CLK_IGNORE_UNUSED should be the handoff type (leaf-clocks used by some component we don't control yet).

The clocks we already register as critical most likely are the ones that should be critical, as they are central clocks where it causes havok when they get disabled :-) .

Heiko

Quoting Heiko Stuebner (2016-02-12 03:39:43)

I =

uld =

OK. It sounds like you might go back and clean up these clocks to use the new infrastructure once it gets merged, so I'll go ahead and pull this tag now.

Thanks,
Mike

Am Freitag, 12. Februar 2016, 07:13:29 schrieb Michael Turquette:

that has always be the plan ;-) .

thanks, that will make life a bit easier when I don't have to rip the branch apart again, but can do the conversion on top.