[v5] reset: amlogic: move audio reset drivers out of CCF

[PATCH v5 0/9] reset: amlogic: move audio reset drivers out of CCF

Posted by Jerome Brunet 2 months, 2 weeks ago

This patchset follows the discussion about having reset driver in the
clock tree [1]. Ideally those should reside in the reset part of tree.

Also the code of the amlogic reset driver is very similar between the 2
trees and could use the same driver code.

This patcheset alignes the reset drivers present in the reset and clock
then adds support for the reset driver of audio clock controller found in
the g12 and sm1 SoC family to the reset tree, using the auxiliary bus.

The infrastructure put in place is meant to be generic enough so we may
eventually also move the reset drivers in the meson8b and aoclk clock
controllers.

This was tested on sm1 vim3l and gxl aml-s905x-cc.

Changes since v4 [6]:
 * Fix Kconfig depends and select as reported by Philipp and
   and 0-day robot.

Changes since v3 [5]:
 * Drop pltf/platform as suggested

Changes since v2 [4]:
 * Fix undefined read access of the reset register
 * Fix Kconfig symbol description

Changes since v1 [3]:
 * Fixes formatting errors reported by Stephen.
 * Changed parameters type to unsigned
 * Fix usage of ops passed as parameters, previously ignored.
 * Return 0 instead of an error if reset support is absent
   to properly decouple from the clock and have a weak
   dependency
 * Split the platform and auxiliary modules in 2 distinct modules
   to fix the COMPILE_TEST error reported by ktest robot.

Change since RFC [2]:
 * Move the aux registration helper out of clock too.

[1] https://lore.kernel.org/linux-clk/e3a85852b911fdf16dd9ae158f42b3ef.sboyd@kernel.org
[2] https://lore.kernel.org/linux-clk/20240516150842.705844-1-jbrunet@baylibre.com
[3] https://lore.kernel.org/linux-clk/20240710162526.2341399-1-jbrunet@baylibre.com
[4] https://lore.kernel.org/linux-clk/20240718095755.3511992-1-jbrunet@baylibre.com
[5] https://lore.kernel.org/linux-clk/20240808102742.4095904-1-jbrunet@baylibre.com
[6] https://lore.kernel.org/r/20240906-meson-rst-aux-v4-0-08824c3d108b@baylibre.com

---
Jerome Brunet (9):
      reset: amlogic: convert driver to regmap
      reset: amlogic: use generic data matching function
      reset: amlogic: make parameters unsigned
      reset: amlogic: add driver parameters
      reset: amlogic: use reset number instead of register count
      reset: amlogic: add reset status support
      reset: amlogic: move drivers to a dedicated directory
      reset: amlogic: split the device core and platform probe
      reset: amlogic: add auxiliary reset driver support

 drivers/reset/Kconfig                              |  15 +-
 drivers/reset/Makefile                             |   3 +-
 drivers/reset/amlogic/Kconfig                      |  27 ++++
 drivers/reset/amlogic/Makefile                     |   4 +
 .../reset/{ => amlogic}/reset-meson-audio-arb.c    |   0
 drivers/reset/amlogic/reset-meson-aux.c            | 136 ++++++++++++++++++
 drivers/reset/amlogic/reset-meson-common.c         | 142 ++++++++++++++++++
 drivers/reset/amlogic/reset-meson.c                | 105 ++++++++++++++
 drivers/reset/amlogic/reset-meson.h                |  28 ++++
 drivers/reset/reset-meson.c                        | 159 ---------------------
 include/soc/amlogic/reset-meson-aux.h              |  23 +++
 11 files changed, 467 insertions(+), 175 deletions(-)
---
base-commit: 487b1b32e317b85c2948eb4013f3e089a0433d49
change-id: 20240906-meson-rst-aux-a72bdc01bd1e

Best regards,
-- 
Jerome

Re: [PATCH v5 0/9] reset: amlogic: move audio reset drivers out of CCF

Posted by Philipp Zabel 2 months, 2 weeks ago

On Di, 2024-09-10 at 18:32 +0200, Jerome Brunet wrote:
> This patchset follows the discussion about having reset driver in the
> clock tree [1]. Ideally those should reside in the reset part of tree.
> 
> Also the code of the amlogic reset driver is very similar between the 2
> trees and could use the same driver code.
> 
> This patcheset alignes the reset drivers present in the reset and clock
> then adds support for the reset driver of audio clock controller found in
> the g12 and sm1 SoC family to the reset tree, using the auxiliary bus.
> 
> The infrastructure put in place is meant to be generic enough so we may
> eventually also move the reset drivers in the meson8b and aoclk clock
> controllers.
> 
> This was tested on sm1 vim3l and gxl aml-s905x-cc.
> 
> Changes since v4 [6]:
>  * Fix Kconfig depends and select as reported by Philipp and
>    and 0-day robot.
> 
> Changes since v3 [5]:
>  * Drop pltf/platform as suggested
> 
> Changes since v2 [4]:
>  * Fix undefined read access of the reset register
>  * Fix Kconfig symbol description
> 
> Changes since v1 [3]:
>  * Fixes formatting errors reported by Stephen.
>  * Changed parameters type to unsigned
>  * Fix usage of ops passed as parameters, previously ignored.
>  * Return 0 instead of an error if reset support is absent
>    to properly decouple from the clock and have a weak
>    dependency
>  * Split the platform and auxiliary modules in 2 distinct modules
>    to fix the COMPILE_TEST error reported by ktest robot.
> 
> Change since RFC [2]:
>  * Move the aux registration helper out of clock too.

Thank you,

Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>

regards
Philipp