Re: [PATCH V5 3/6] mfd: pm8008: Add mfd cell struct to register LDOs

From: Satya Priya Kakitapalli (Temp)
Date: Fri Feb 11 2022 - 05:12:49 EST



On 2/10/2022 7:02 AM, Stephen Boyd wrote:
Quoting Satya Priya (2022-02-08 06:52:17)
diff --git a/drivers/mfd/qcom-pm8008.c b/drivers/mfd/qcom-pm8008.c
index c472d7f..e8569cc 100644
--- a/drivers/mfd/qcom-pm8008.c
+++ b/drivers/mfd/qcom-pm8008.c
@@ -8,6 +8,7 @@
#include <linux/interrupt.h>
#include <linux/irq.h>
#include <linux/irqdomain.h>
+#include <linux/mfd/core.h>
#include <linux/module.h>
#include <linux/of_device.h>
#include <linux/of_platform.h>
@@ -27,6 +28,37 @@
#define INT_EN_CLR_OFFSET 0x16
#define INT_LATCHED_STS_OFFSET 0x18

+static const struct mfd_cell pm8008_regulator_devs[] = {
Is there some way to not allocate this structure statically forever?


I think No.

I found that some of the drivers are just using one cell with .name to match with regulator driver and then probing regulators using a loop. I'll do that too.

static const struct mfd_cell pm8008_regulator_devs[] = {
        {
                .name = "qcom,pm8008-regulators",
        },
 };

+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 0,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 1,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 2,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 3,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 4,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 5,
+ },
+ {
+ .name = "qcom,pm8008-regulators",
+ .id = 6,
+ },
+};
+
enum {
PM8008_MISC,
pm8008_temp_alarm,
@@ -35,6 +67,17 @@ enum {
PM8008_NUM_PERIPHS,
};

+enum {
+ PM8008_INFRA,
+ PM8008_REGULATORS,
+};
+
+static const struct of_device_id pm8008_match[] = {
+ { .compatible = "qcom,pm8008", .data = (void *)PM8008_INFRA},
+ { .compatible = "qcom,pm8008-regulators", .data = (void *)PM8008_REGULATORS},
+ { },
Nitpick: Drop , on {} so nothing can come after without causing compile
error.


Okay.


+};
+
#define PM8008_PERIPH_0_BASE 0x900
#define PM8008_PERIPH_1_BASE 0x2400
#define PM8008_PERIPH_2_BASE 0xC000
@@ -221,6 +264,7 @@ static int pm8008_probe(struct i2c_client *client)
{
int rc;
struct pm8008_data *chip;
+ const struct of_device_id *id;

chip = devm_kzalloc(&client->dev, sizeof(*chip), GFP_KERNEL);
if (!chip)
@@ -239,14 +283,19 @@ static int pm8008_probe(struct i2c_client *client)
dev_err(chip->dev, "Failed to probe irq periphs: %d\n", rc);
}

+ id = of_match_node(pm8008_match, chip->dev->of_node);
Use device_get_match_data()? And then use (uintptr_t) casts to check for
the enum? Using device_get_match_data() allows us to avoid moving the
pm8008_match table.


Okay.


+ if (id->data == (void *)PM8008_REGULATORS) {
enum <your_name_here> dev_type = device_get_match_data(dev);

if (dev_type == PM8008_REGULATORS)

+ rc = mfd_add_devices(chip->dev, 0, pm8008_regulator_devs,
Why not devm_mfd_add_devices()?


Okay.


+ ARRAY_SIZE(pm8008_regulator_devs), NULL, 0, NULL);
+ if (rc) {
+ dev_err(chip->dev, "Failed to add children: %d\n", rc);
+ return rc;
+ }
+ }
+
return devm_of_platform_populate(chip->dev);
}

-static const struct of_device_id pm8008_match[] = {
- { .compatible = "qcom,pm8008", },
- { },
-};
This should have a MODULE_DEVICE_TABLE(of, pm8008_match) here.


Okay.


-
static struct i2c_driver pm8008_mfd_driver = {
.driver = {
.name = "pm8008",