[v2,0/3] Add CRU support for rk3576 SoC

Message

Detlev Casanova Aug. 2, 2024, 9:35 p.m. UTC

Add support for clocks and resets on the rk3576.
Patches from downstream have been squashed and rebased.

The resets have been renumbered without gaps and their actual register/bit
information is set in rst-rk3576.c as it has been done for rk3588.

Changes since v1:
- Remove reset defines that are probably out of the main core
- Separate resets and clocks bindings
- Renumber the resets without gaps

Detlev.

Detlev Casanova (1):
  dt-bindings: clock: add rk3576 cru bindings

Elaine Zhang (2):
  clk: rockchip: Add dt-binding header for rk3576
  clk: rockchip: Add clock controller for the RK3576

 .../bindings/clock/rockchip,rk3576-cru.yaml   |   73 +
 drivers/clk/rockchip/Kconfig                  |    7 +
 drivers/clk/rockchip/Makefile                 |    1 +
 drivers/clk/rockchip/clk-rk3576.c             | 1819 +++++++++++++++++
 drivers/clk/rockchip/clk.h                    |   53 +
 drivers/clk/rockchip/rst-rk3576.c             |  555 +++++
 .../dt-bindings/clock/rockchip,rk3576-cru.h   |  589 ++++++
 .../dt-bindings/reset/rockchip,rk3576-cru.h   |  484 +++++
 8 files changed, 3581 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/clock/rockchip,rk3576-cru.yaml
 create mode 100644 drivers/clk/rockchip/clk-rk3576.c
 create mode 100644 drivers/clk/rockchip/rst-rk3576.c
 create mode 100644 include/dt-bindings/clock/rockchip,rk3576-cru.h
 create mode 100644 include/dt-bindings/reset/rockchip,rk3576-cru.h

Comments

Heiko Stübner Aug. 6, 2024, 3:13 p.m. UTC | #1

Hi Detlev,

Am Dienstag, 6. August 2024, 16:15:41 CEST schrieb Detlev Casanova:
> The suggestion from Heiko was that those reset should be managed by the 
> subsystems that use them, because they are on a different offset and therefore 
> seem to be on a different core.
> 
> But I think I will include them here like you suggested because:
>  - That's actually how it is done for rk3588 (which is quite close th rk3576),
>  - According to you and the TRM, those resets are on the same core, just with 
> big offsets.
> 
> Having the same structure for both SoC makes sense for maintening them.

Just without the big offsets between areas please.
Similar to how rk3588 does it already.

And yep most likely they are in the same block. Just that huge block of
space for the cru somehow suggested some algamation of multiple ones,
but looking up the rk3588, you're right that it really seems to be one block.

I did request the rk3576 TRM from Rockchip - hopefully they'll follow up
with that at some point ;-) .


Heiko