mbox series

[PATCHv7,00/11] Introduce RK806 Support

Message ID 20230307153617.643260-1-sebastian.reichel@collabora.com (mailing list archive)
Headers show
Series Introduce RK806 Support | expand

Message

Sebastian Reichel March 7, 2023, 3:36 p.m. UTC
Hi,

The Rockchip RK3588 Evaluation Boards as well as the Rock 5
boards and probably most other RK3588 boards use SPI connected
RK806 PMICs. Downstream this is handled by a new driver, but
apart from being SPI connected this chip is quite similar to
the other Rockchip PMICs (and RK806 is promoted to also support
I2C). Thus this series instead updates the RK808 driver(s).

Changelog since PATCHv6:
 * https://lore.kernel.org/all/20230127181244.160887-1-sebastian.reichel@collabora.com/
 * rebased to v6.3-rc1
 * add Lee Jones' Acked-by-for-MFD to patches 5 & 6
 * add Linus Walleij Acked-by to patch 9
 * fix review comments provided by Lee Jones
   - remove spaces around ## in RK806_CMD_WITH_SIZE
   - add spaces around - in RK806_CMD_WITH_SIZE
   - change regmap init error message
   - further improve readability of buffer handling in rk806_spi_bus_write
   - use define for maximum buffer size
 * add .max_register to regmap_config, so that they can be analyzed in debugfs
 * switch to 16 bit register addresses to further simplify the code
 * fix max. buffer size by using RK806_CMD_LEN_MSK,
   which is 4 bit (max. 16 bytes) instead of 7 bit (max. 128 bytes)
 * add RK806_CMD_CRC_DIS to RK806_CMD_WITH_SIZE define to explicitly
   initialize all bits of the command byte

Changelog since PATCHv5:
 * https://lore.kernel.org/all/20230109172723.60304-1-sebastian.reichel@collabora.com/
 * add Reviewed-by from Rob Herring to the rk806 DT binding patch
 * drop useless wrapper functions for rk8xx_i2c_suspend and rk8xx_i2c_resume
 * put .num_resources after .resources for rk806_pwrkey_resources
 * simplify dual_support logic by storing IRQF_SHARED or 0 instead of bool
 * add missing spaces around + and - in rk806
 * add new patch using device_get_match_data() in rk8xx-i2c and removing
   the pointless dev_info printing the chip version
 * call devm_mfd_add_devices() with id=0 and rely on mfd_cell to have the .id
   field configured to either PLATFORM_DEVID_AUTO or PLATFORM_DEVID_NONE
 * add check if more than 128 bytes are tried to be received/sent, which would
   overflow the command register
 * add RK806_CMD_WITH_SIZE() define

Changelog since PATCHv4:
 * https://lore.kernel.org/all/20221020204251.108565-1-sebastian.reichel@collabora.com/
 * rebase to v6.2-rc1
   - dropped 'regulator: rk808: reduce 'struct rk808' usage' (queued to 6.2-rc1)
   - dropped 'regulator: rk808: Use dev_err_probe' (queued to 6.2-rc1)
   - dropped 'rtc: rk808: reduce 'struct rk808' usage' (queued to 6.2-rc1)
 * use 'MFD_RK8XX' for the MFD driver supporting all the MFD8XX chips
 * added author tags to drivers/mfd/rk8xx-core.c. They were missing because
   I moved the original header over to the i2c specific file and wrote the
   new header from scratch. I suppose it's better to have the author tags
   in both files.
 * fix Rob's comments for the rk806 YAML binding
 * add defines for rk806 command indexes
 * modify rk806 code, so that it is capable of mult-write

Changelog since PATCHv3:
 * https://lore.kernel.org/all/20220909175522.179175-1-sebastian.reichel@collabora.com/
 * Dropped removing REGMAP_I2C dependency from RK817 ASoC driver (applied)
 * Rename MFD_RK808 to MFD_RK8XX to be consistent. It makes sense to do this now,
   since the patchset touches all the child drivers anyways.
 * rebase to v6.1-rc1
 * collected a couple of Acks
 * update rk806 DT binding according to DT maintainer feedback
 * add missing pinmux config to the rk806 DT binding
 * update rk806_spi_bus_write and rk806_spi_bus_read
 * replaced some constants with sizeof or defines
 * used capitalized comments
 * rename regmap_find_closest_bigger to regulator_find_closest_bigger, not sure
   why I prefixed it with regmap_ in the first place
 * use rk8xx_is_enabled_wmsk_regmap instead of regulator_is_enabled_regmap for
   the switching regulators to correctly report the state
 * reordered the first few patches grouping the MFD patches together

Changelog since PATCHv2:
 * https://lore.kernel.org/all/20220908003107.220143-1-sebastian.reichel@collabora.com/
 * Change DT binding to not allow nldo-reg6
 * Fix DT binding to check for [np]ldo-reg instead of [np]ldo_reg
 * remove rk806_get_voltage_sel_regmap in favour of regulator_get_voltage_sel_regmap
 * drop rk806_set_voltage in favour of regulator_set_voltage_sel_regmap
 * use regulator_set_ramp_delay_regmap
 * drop possibly incorrect printing of chip id register address in case of errors

Changelog since PATCHv1:
 * https://lore.kernel.org/all/20220831215437.117880-1-sebastian.reichel@collabora.com/
 * Collect Acked-by
 * Avoid if/else checks for regulator id in rk806 regulator driver
 * Fix indentation in DTS example section of the rk806 binding
 * Use absolute path for regulator.yaml referencing in the rk806 binding
 * Reduce pattern for DCDC regulators to only allow 1-10
 * replace uppercase name with lowercase ones in regulator names
 * replace _ with - in regulator names

-- Sebastian

Sebastian Reichel (11):
  clk: RK808: reduce 'struct rk808' usage
  mfd: rk808: convert to device managed resources
  mfd: rk808: use dev_err_probe
  mfd: rk808: replace 'struct i2c_client' with 'struct device'
  mfd: rk808: split into core and i2c
  mfd: rk8xx-i2c: use device_get_match_data
  dt-bindings: mfd: add rk806 binding
  mfd: rk8xx: add rk806 support
  pinctrl: rk805: add rk806 pinctrl support
  regulator: expose regulator_find_closest_bigger
  regulator: rk808: add rk806 support

 .../bindings/mfd/rockchip,rk806.yaml          | 406 +++++++++++++++++
 drivers/clk/Kconfig                           |   2 +-
 drivers/clk/clk-rk808.c                       |  34 +-
 drivers/input/misc/Kconfig                    |   2 +-
 drivers/mfd/Kconfig                           |  21 +-
 drivers/mfd/Makefile                          |   4 +-
 drivers/mfd/{rk808.c => rk8xx-core.c}         | 352 +++++----------
 drivers/mfd/rk8xx-i2c.c                       | 185 ++++++++
 drivers/mfd/rk8xx-spi.c                       | 124 ++++++
 drivers/pinctrl/Kconfig                       |   2 +-
 drivers/pinctrl/pinctrl-rk805.c               | 189 +++++++-
 drivers/power/supply/Kconfig                  |   2 +-
 drivers/regulator/Kconfig                     |   2 +-
 drivers/regulator/helpers.c                   |  22 +-
 drivers/regulator/rk808-regulator.c           | 383 ++++++++++++++++
 drivers/rtc/Kconfig                           |   2 +-
 include/linux/mfd/rk808.h                     | 417 +++++++++++++++++-
 include/linux/regulator/driver.h              |   2 +
 sound/soc/codecs/Kconfig                      |   2 +-
 19 files changed, 1867 insertions(+), 286 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk806.yaml
 rename drivers/mfd/{rk808.c => rk8xx-core.c} (71%)
 create mode 100644 drivers/mfd/rk8xx-i2c.c
 create mode 100644 drivers/mfd/rk8xx-spi.c

Comments

Diederik de Haas April 26, 2023, 3:20 p.m. UTC | #1
On Tuesday, 7 March 2023 16:36:06 CEST Sebastian Reichel wrote:
> The Rockchip RK3588 Evaluation Boards as well as the Rock 5
> boards and probably most other RK3588 boards use SPI connected
> RK806 PMICs. Downstream this is handled by a new driver, but
> apart from being SPI connected this chip is quite similar to
> the other Rockchip PMICs (and RK806 is promoted to also support
> I2C). Thus this series instead updates the RK808 driver(s).

I don't know if it's relevant or useful, but I tested whether this patch set 
would cause a regression on the Rock64 (rk3328), Quartz64 Model A and Quartz64 
Model B (both rk3566) and I can say that it does NOT cause a regression.

If it is relevant or useful, feel free to add my 
Tested-by: Diederik de Haas <didi.debian@cknow.org>