On 13/10/2021 23:39, Luca Ceresoli wrote:

Hi,

On 12/10/21 10:32, Krzysztof Kozlowski wrote:

quoted

On 11/10/2021 17:56, Luca Ceresoli wrote:

quoted

Add a simple driver for the Maxim MAX77714 PMIC, supporting RTC and
watchdog only.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
---
 MAINTAINERS                  |   2 +
 drivers/mfd/Kconfig          |  14 ++++
 drivers/mfd/Makefile         |   1 +
 drivers/mfd/max77714.c       | 151 +++++++++++++++++++++++++++++++++++
 include/linux/mfd/max77714.h |  68 ++++++++++++++++
 5 files changed, 236 insertions(+)
 create mode 100644 drivers/mfd/max77714.c
 create mode 100644 include/linux/mfd/max77714.h

(...)

quoted

+
+static const struct of_device_id max77714_dt_match[] = {
+	{ .compatible = "maxim,max77714" },
+	{},
+};

When converting to module - don't forget the MODULE_DEVICE_TABLE

quoted

+
+static struct i2c_driver max77714_driver = {
+	.driver = {
+		.name = "max77714",
+		.of_match_table = of_match_ptr(max77714_dt_match),

Kbuild robot pointed it out - of_matc_ptr should not be needed, even for
compile testing without OF.

I wonder whether it's better to add '#ifdef CONFIG_OF / #endif' around
the struct of_device_id declaration. I think it's what most drivers do,
even though I tend to prefer not adding #ifdefs making code less clean
only for COMPILE_TESTING.

No, most drivers added it long time ago before we switched it to a new
way - either __maybe_unused or without anything even. The point is that
OF driver can be reused for ACPI platforms. If you limit it with ifdef
or of_match_ptr, the ACPI platform won't have any table to use for binding.

Best regards,
Krzysztof