[permalink] [raw]

Subject: Re: [PATCHv2 12/13] regulator: rk808: add rk806 support

On 9/8/22 03:31, Sebastian Reichel wrote:
> Add rk806 support to the existing rk808 regulator
> driver.
>
> This has been implemented using shengfei Xu's rk806
> specific driver from the vendor tree as reference.
>
> Co-Developed-by: shengfei Xu <[email protected]>
> Signed-off-by: shengfei Xu <[email protected]>
> Signed-off-by: Sebastian Reichel <[email protected]>
> ---
> drivers/regulator/rk808-regulator.c | 482 ++++++++++++++++++++++++++++
> 1 file changed, 482 insertions(+)
>
> diff --git a/drivers/regulator/rk808-regulator.c b/drivers/regulator/rk808-regulator.c
> index fa9fc1aa1ae3..cd1a2cff4a37 100644
> --- a/drivers/regulator/rk808-regulator.c
> +++ b/drivers/regulator/rk808-regulator.c

Thanks for upstreaming the downstream stuff! :)
I wonder if we could drop some code by using the existing helpers? Or
maybe I am missreading some code. Wouldn't be the first (and probably
not the last) time...

//snip

> struct rk808_regulator_data {
> struct gpio_desc *dvs_gpio[2];
> };
> @@ -216,6 +271,223 @@ static const unsigned int rk817_buck1_4_ramp_table[] = {
> 3000, 6300, 12500, 25000
> };
>
> +static int rk806_get_voltage_sel_regmap(struct regulator_dev *rdev)
> +{
> + unsigned int val;
> + int vsel_reg, ret;
> +
> + vsel_reg = rdev->desc->vsel_reg;
> +
> + ret = regmap_read(rdev->regmap, vsel_reg, &val);
> + if (ret != 0)
> + return ret;
> +
> + val &= rdev->desc->vsel_mask;
> + val >>= ffs(rdev->desc->vsel_mask) - 1;
> +
> + return val;
> +}

Could we just use the regulator_get_voltage_sel_regmap()?

> +
> +static int rk806_set_voltage(struct regulator_dev *rdev,
> + int req_min_uV, int req_max_uV,
> + unsigned int *selector)
> +{
> + int vsel_reg, ret, sel;
> +
> + ret = regulator_map_voltage_linear_range(rdev, req_min_uV, req_max_uV);
> + if (ret >= 0) {
> + *selector = ret;
> + sel = ret;
> + } else {
> + return -EINVAL;
> + }
> +
> + vsel_reg = rdev->desc->vsel_reg;
> +
> + sel <<= ffs(rdev->desc->vsel_mask) - 1;
> +
> + ret = regmap_update_bits(rdev->regmap, vsel_reg,
> + rdev->desc->vsel_mask, sel);
> +
> + return ret;
> +}

Hmm. Maybe this is not necessary? I wonder if we could get away just
with the .map and .set_voltage_sel (regulator_get_voltage_sel_regmap()
and regulator_map_voltage_linear_range()).

// snip
> +
> +static int rk806_set_ramp_delay_ldo(struct regulator_dev *rdev, int ramp_delay)
> +{
> + unsigned int ramp_value = RK806_RAMP_RATE_2LSB_PER_1CLK;
> + int regval;
> +
> + switch (ramp_delay) {
> + case 1 ... 780:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_32CLK;
> + break;
> + case 781 ... 1900:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_13CLK;
> + break;
> + case 1901 ... 3120:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_8CLK;
> + break;
> + case 3121 ... 6280:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_4CLK;
> + break;
> + case 6281 ... 12500:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_2CLK;
> + break;
> + case 12501 ... 25000:
> + ramp_value = RK806_RAMP_RATE_1LSB_PER_1CLK;
> + break;
> + case 25001 ... 50000:
> + ramp_value = RK806_RAMP_RATE_2LSB_PER_1CLK;
> + break;
> + case 50001 ... 100000:
> + ramp_value = RK806_RAMP_RATE_4LSB_PER_1CLK;
> + break;
> + default:
> + pr_warn("%s ramp_delay: %d not supported, setting 10000\n",
> + rdev->desc->name, ramp_delay);
> + }
> +
> + regval = ramp_value << (ffs(rdev->desc->ramp_mask) - 1);
> + return regmap_update_bits(rdev->regmap, rdev->desc->ramp_reg,
> + rdev->desc->ramp_mask, regval);
> +}

Do you think we could get rid of this function by populating a
ramp-delay table and using regulator_set_ramp_delay_regmap()?

Best Regards
-- Matti

--
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland

~~ When things go utterly wrong vim users can always type :help! ~~

2022-09-08 09:19:35

by Krzysztof Kozlowski

[permalink] [raw]

Subject: Re: [PATCHv2 08/13] dt-bindings: mfd: add rk806 binding

On 08/09/2022 02:31, Sebastian Reichel wrote:
> Add DT binding document for Rockchip's RK806 PMIC.
>
> Signed-off-by: Sebastian Reichel <[email protected]>
> ---

Thank you for your patch. There is something to discuss/improve.

> + vcca-supply:
> + description:
> + The input supply for pldo-reg6.
> +
> + regulators:
> + type: object
> + patternProperties:
> + "^(dcdc-reg([1-9]|10)|[pn]ldo_reg[1-6])$":

You still have underscore in the pattern. This should pop-up during
test: please run `make dt_binding_check` (see
Documentation/devicetree/bindings/writing-schema.rst for instructions).

Best regards,
Krzysztof

2022-09-08 09:21:12

by Robin Murphy

[permalink] [raw]

Subject: Re: [PATCHv2 02/13] mfd: rk808: use dev_err_probe

On 2022-09-08 01:30, Sebastian Reichel wrote:
> Use dev_err_probe instead of dev_err in probe function,
> which simplifies code a little bit and prints the error
> code.
>
> Signed-off-by: Sebastian Reichel <[email protected]>
> ---
> drivers/mfd/rk808.c | 50 ++++++++++++++++-----------------------------
> 1 file changed, 18 insertions(+), 32 deletions(-)
>
> diff --git a/drivers/mfd/rk808.c b/drivers/mfd/rk808.c
> index c17fea1d48ca..e793de9146f7 100644
> --- a/drivers/mfd/rk808.c
> +++ b/drivers/mfd/rk808.c
> @@ -650,18 +650,14 @@ static int rk808_probe(struct i2c_client *client,
>
> /* Read chip variant */
> msb = i2c_smbus_read_byte_data(client, pmic_id_msb);
> - if (msb < 0) {
> - dev_err(&client->dev, "failed to read the chip id at 0x%x\n",
> - RK808_ID_MSB);
> - return msb;
> - }
> + if (msb < 0)
> + return dev_err_probe(&client->dev, msb, "failed to read the chip id at 0x%x\n",
> + RK808_ID_MSB);
>
> lsb = i2c_smbus_read_byte_data(client, pmic_id_lsb);
> - if (lsb < 0) {
> - dev_err(&client->dev, "failed to read the chip id at 0x%x\n",
> - RK808_ID_LSB);
> - return lsb;
> - }
> + if (lsb < 0)
> + return dev_err_probe(&client->dev, lsb, "failed to read the chip id at 0x%x\n",
> + RK808_ID_LSB);

Hmm, refactoring these is probably a good opportunity to fix them as
well - I'm not sure printing the register offset is really all that
useful, but particularly not when it's hard-coded and may not correspond
to the actual access that failed.

Robin.

>
> rk808->variant = ((msb << 8) | lsb) & RK8XX_ID_MSK;
> dev_info(&client->dev, "chip id: 0x%x\n", (unsigned int)rk808->variant);
> @@ -710,44 +706,34 @@ static int rk808_probe(struct i2c_client *client,
> i2c_set_clientdata(client, rk808);
>
> rk808->regmap = devm_regmap_init_i2c(client, rk808->regmap_cfg);
> - if (IS_ERR(rk808->regmap)) {
> - dev_err(&client->dev, "regmap initialization failed\n");
> - return PTR_ERR(rk808->regmap);
> - }
> + if (IS_ERR(rk808->regmap))
> + return dev_err_probe(&client->dev, PTR_ERR(rk808->regmap),
> + "regmap initialization failed\n");
>
> - if (!client->irq) {
> - dev_err(&client->dev, "No interrupt support, no core IRQ\n");
> - return -EINVAL;
> - }
> + if (!client->irq)
> + return dev_err_probe(&client->dev, -EINVAL, "No interrupt support, no core IRQ\n");
>
> ret = devm_regmap_add_irq_chip(&client->dev, rk808->regmap, client->irq,
> IRQF_ONESHOT, -1,
> rk808->regmap_irq_chip, &rk808->irq_data);
> - if (ret) {
> - dev_err(&client->dev, "Failed to add irq_chip %d\n", ret);
> - return ret;
> - }
> + if (ret)
> + return dev_err_probe(&client->dev, ret, "Failed to add irq_chip\n");
>
> for (i = 0; i < nr_pre_init_regs; i++) {
> ret = regmap_update_bits(rk808->regmap,
> pre_init_reg[i].addr,
> pre_init_reg[i].mask,
> pre_init_reg[i].value);
> - if (ret) {
> - dev_err(&client->dev,
> - "0x%x write err\n",
> - pre_init_reg[i].addr);
> - return ret;
> - }
> + if (ret)
> + return dev_err_probe(&client->dev, ret, "0x%x write err\n",
> + pre_init_reg[i].addr);
> }
>
> ret = devm_mfd_add_devices(&client->dev, PLATFORM_DEVID_NONE,
> cells, nr_cells, NULL, 0,
> regmap_irq_get_domain(rk808->irq_data));
> - if (ret) {
> - dev_err(&client->dev, "failed to add MFD devices %d\n", ret);
> - return ret;
> - }
> + if (ret)
> + return dev_err_probe(&client->dev, ret, "failed to add MFD devices\n");
>
> if (of_property_read_bool(np, "rockchip,system-power-controller")) {
> ret = devm_register_sys_off_handler(&client->dev,

2022-09-08 11:38:02

by Mark Brown

[permalink] [raw]

Subject: Re: [PATCHv2 12/13] regulator: rk808: add rk806 support

On Thu, Sep 08, 2022 at 10:45:40AM +0300, Matti Vaittinen wrote:
> On 9/8/22 03:31, Sebastian Reichel wrote:

> > +static int rk806_set_voltage(struct regulator_dev *rdev,
> > + int req_min_uV, int req_max_uV,
> > + unsigned int *selector)

> Hmm. Maybe this is not necessary? I wonder if we could get away just with
> the .map and .set_voltage_sel (regulator_get_voltage_sel_regmap() and
> regulator_map_voltage_linear_range()).

Not just can, should. Drivers should not mix and match voltage based
and selector based operations, and they should always try to implement
selector based operations unless there's a specific reason to do
otherwise.

Attachments:

(No filename) (677.00 B)
signature.asc (499.00 B)
Download all attachments

2022-09-08 13:24:04

by Rob Herring (Arm)

[permalink] [raw]

Subject: Re: [PATCHv2 08/13] dt-bindings: mfd: add rk806 binding

On Thu, 08 Sep 2022 02:31:02 +0200, Sebastian Reichel wrote:
> Add DT binding document for Rockchip's RK806 PMIC.
>
> Signed-off-by: Sebastian Reichel <[email protected]>
> ---
> .../bindings/mfd/rockchip,rk806.yaml | 388 ++++++++++++++++++
> 1 file changed, 388 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk806.yaml
>

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check'
on your patch (DT_CHECKER_FLAGS is new in v5.13):

yamllint warnings/errors:

dtschema/dtc warnings/errors:
/builds/robherring/dt-review-ci/linux/Documentation/devicetree/bindings/mfd/rockchip,rk806.example.dtb: pmic@0: regulators: Unevaluated properties are not allowed ('nldo-reg1', 'nldo-reg2', 'nldo-reg3', 'nldo-reg4', 'nldo-reg5', 'pldo-reg1', 'pldo-reg2', 'pldo-reg3', 'pldo-reg4', 'pldo-reg5', 'pldo-reg6' were unexpected)
From schema: /builds/robherring/dt-review-ci/linux/Documentation/devicetree/bindings/mfd/rockchip,rk806.yaml

doc reference errors (make refcheckdocs):

See https://patchwork.ozlabs.org/patch/

This check can fail if there are any dependencies. The base for a patch
series is generally the most recent rc1.

If you already ran 'make dt_binding_check' and didn't see the above
error(s), then make sure 'yamllint' is installed and dt-schema is up to
date:

pip3 install dtschema --upgrade

Please check and re-submit.