[PATCH RFC 0/4] Start unification of of_device_id and i2c_device_id

[PATCH RFC 0/4] Start unification of of_device_id and i2c_device_id

[Dmitry Torokhov]

i2c_device_id is using unsigned long as driver data, whereas
of_device_id is using a void pointer. Normally drivers use them both as
either a pointer or a scalar, but in some cases driver's authors chose
to treat one as a pointer and another as a scalar that is being used as
an index in array of chip-specific data, i.e. something like this:

	data = device_get_match_data(...);
	if (!data) {
		id = i2c_client_get_device_id(...);
		data = &some_array[id->driver_data];
	}

This makes it hard to turn device_get_match_data() into truly generic
function working well for ACPI and DT, as well as for legacy cases.

So let's try to switch i2c_device_id to also use a void pointer and
first switch drivers with the patterns above to it, and then the rest of
them.

I did a conversion of a couple of drivers in i2c to illustrate.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
Dmitry Torokhov (4):
      i2c: start migrating to a pointer in i2c_device_id
      i2c: mux: ltc4306: convert to using a pointer in i2c_device_id
      i2c: mux: pca954x: convert to using a pointer in i2c_device_id
      i2c: slave-eeprom: convert to using a pointer in i2c_device_id

 drivers/i2c/i2c-core-base.c         |  2 +-
 drivers/i2c/i2c-slave-eeprom.c      | 99 ++++++++++++++++++++++++-------------
 drivers/i2c/muxes/i2c-mux-ltc4306.c |  7 ++-
 drivers/i2c/muxes/i2c-mux-pca954x.c | 28 +++++------
 include/linux/mod_devicetable.h     |  3 +-
 5 files changed, 86 insertions(+), 53 deletions(-)
---
base-commit: 21ef7b1e17d039053edaeaf41142423810572741
change-id: 20230814-i2c-id-rework-9a4742e9ee52

Best regards,
-- 
Dmitry

[PATCH RFC 1/4] i2c: start migrating to a pointer in i2c_device_id

[Dmitry Torokhov]

The of_device_id structure uses a void pointer to associate additional
driver-private data with device id, most commonly used to refer to a
structure containing additional characteristics of a particular chip.
However i2c_device_id uses an unsigned long. While it can easily be
converted to a pointer, several drivers use it to store a scalar (and it
is possible to use a pointer in of_device_id to store a scalar value as
well). The worst case comes when OF part of a driver uses pointers,
while legacy part is using scalars, causing constructs like:

	data = device_get_match_data(...);
	if (!data)
		data = &some_array[i2c_match_id(...)->data];
	...

To avoid this introduce a const void "data" pointer to i2c_device_id as
well, so that we can convert drivers one by one, and drop driver_data
member in the end.

The end goal is to clean up all helpers and make device_get_match_data()
work universally for all ACPI, DT, and legacy variants.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/i2c/i2c-core-base.c     | 2 +-
 include/linux/mod_devicetable.h | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
index 60746652fd52..75d12d79d8f1 100644
--- a/drivers/i2c/i2c-core-base.c
+++ b/drivers/i2c/i2c-core-base.c
@@ -126,7 +126,7 @@ const void *i2c_get_match_data(const struct i2c_client *client)
 		if (!match)
 			return NULL;
 
-		data = (const void *)match->driver_data;
+		data = match->data ?: (const void *)match->driver_data;
 	}
 
 	return data;
diff --git a/include/linux/mod_devicetable.h b/include/linux/mod_devicetable.h
index b0678b093cb2..21ac2a96e828 100644
--- a/include/linux/mod_devicetable.h
+++ b/include/linux/mod_devicetable.h
@@ -477,7 +477,8 @@ struct rpmsg_device_id {
 
 struct i2c_device_id {
 	char name[I2C_NAME_SIZE];
-	kernel_ulong_t driver_data;	/* Data private to the driver */
+	kernel_ulong_t driver_data;	/* This field is obsolete */
+	const void *data;		/* Data private to the driver */
 };
 
 /* pci_epf */

-- 
2.41.0.694.ge786442a9b-goog

[PATCH RFC 2/4] i2c: mux: ltc4306: convert to using a pointer in i2c_device_id

[Dmitry Torokhov]

Switch the driver to use newly added "data" pointer in i2c_device_id to
streamline DT and legacy flows.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/i2c/muxes/i2c-mux-ltc4306.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/i2c/muxes/i2c-mux-ltc4306.c b/drivers/i2c/muxes/i2c-mux-ltc4306.c
index 637e25506490..17b9ed9ceef3 100644
--- a/drivers/i2c/muxes/i2c-mux-ltc4306.c
+++ b/drivers/i2c/muxes/i2c-mux-ltc4306.c
@@ -192,8 +192,8 @@ static int ltc4306_deselect_mux(struct i2c_mux_core *muxc, u32 chan)
 }
 
 static const struct i2c_device_id ltc4306_id[] = {
-	{ "ltc4305", ltc_4305 },
-	{ "ltc4306", ltc_4306 },
+	{ "ltc4305", .data = &chips[ltc_4305] },
+	{ "ltc4306", .data = &chips[ltc_4306] },
 	{ }
 };
 MODULE_DEVICE_TABLE(i2c, ltc4306_id);
@@ -217,9 +217,8 @@ static int ltc4306_probe(struct i2c_client *client)
 	int num, ret;
 
 	chip = of_device_get_match_data(&client->dev);
-
 	if (!chip)
-		chip = &chips[i2c_match_id(ltc4306_id, client)->driver_data];
+		chip = i2c_match_id(ltc4306_id, client)->data;
 
 	idle_disc = device_property_read_bool(&client->dev,
 					      "i2c-mux-idle-disconnect");

-- 
2.41.0.694.ge786442a9b-goog

[PATCH RFC 3/4] i2c: mux: pca954x: convert to using a pointer in i2c_device_id

[Dmitry Torokhov]

Switch the driver to use newly added "data" pointer in i2c_device_id to
streamline DT and legacy flows.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/i2c/muxes/i2c-mux-pca954x.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/drivers/i2c/muxes/i2c-mux-pca954x.c b/drivers/i2c/muxes/i2c-mux-pca954x.c
index 6965bf4c2348..c8540e4ca660 100644
--- a/drivers/i2c/muxes/i2c-mux-pca954x.c
+++ b/drivers/i2c/muxes/i2c-mux-pca954x.c
@@ -177,18 +177,18 @@ static const struct chip_desc chips[] = {
 };
 
 static const struct i2c_device_id pca954x_id[] = {
-	{ "pca9540", pca_9540 },
-	{ "pca9542", pca_9542 },
-	{ "pca9543", pca_9543 },
-	{ "pca9544", pca_9544 },
-	{ "pca9545", pca_9545 },
-	{ "pca9546", pca_9546 },
-	{ "pca9547", pca_9547 },
-	{ "pca9548", pca_9548 },
-	{ "pca9846", pca_9846 },
-	{ "pca9847", pca_9847 },
-	{ "pca9848", pca_9848 },
-	{ "pca9849", pca_9849 },
+	{ "pca9540", .data = &chips[pca_9540] },
+	{ "pca9542", .data = &chips[pca_9542] },
+	{ "pca9543", .data = &chips[pca_9543] },
+	{ "pca9544", .data = &chips[pca_9544] },
+	{ "pca9545", .data = &chips[pca_9545] },
+	{ "pca9546", .data = &chips[pca_9546] },
+	{ "pca9547", .data = &chips[pca_9547] },
+	{ "pca9548", .data = &chips[pca_9548] },
+	{ "pca9846", .data = &chips[pca_9846] },
+	{ "pca9847", .data = &chips[pca_9847] },
+	{ "pca9848", .data = &chips[pca_9848] },
+	{ "pca9849", .data = &chips[pca_9849] },
 	{ }
 };
 MODULE_DEVICE_TABLE(i2c, pca954x_id);
@@ -206,7 +206,7 @@ static const struct of_device_id pca954x_of_match[] = {
 	{ .compatible = "nxp,pca9847", .data = &chips[pca_9847] },
 	{ .compatible = "nxp,pca9848", .data = &chips[pca_9848] },
 	{ .compatible = "nxp,pca9849", .data = &chips[pca_9849] },
-	{}
+	{ }
 };
 MODULE_DEVICE_TABLE(of, pca954x_of_match);
 
@@ -447,7 +447,7 @@ static int pca954x_probe(struct i2c_client *client)
 
 	data->chip = device_get_match_data(dev);
 	if (!data->chip)
-		data->chip = &chips[id->driver_data];
+		data->chip = id->data;
 
 	if (data->chip->id.manufacturer_id != I2C_DEVICE_ID_NONE) {
 		struct i2c_device_identity id;

-- 
2.41.0.694.ge786442a9b-goog

[PATCH RFC 4/4] i2c: slave-eeprom: convert to using a pointer in i2c_device_id

[Dmitry Torokhov]

Switch the driver to use newly added "data" pointer in i2c_device_id and
introduce struct eeprom_chip to describe chip's characteristics instead
of cramming it all into an unisigned long and then decipher.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/i2c/i2c-slave-eeprom.c | 99 ++++++++++++++++++++++++++++--------------
 1 file changed, 66 insertions(+), 33 deletions(-)

diff --git a/drivers/i2c/i2c-slave-eeprom.c b/drivers/i2c/i2c-slave-eeprom.c
index 5946c0d0aef9..1d2340da5a70 100644
--- a/drivers/i2c/i2c-slave-eeprom.c
+++ b/drivers/i2c/i2c-slave-eeprom.c
@@ -16,32 +16,29 @@
  * EEPROMs, though. We currently use the 8 bit as a valid address.
  */
 
-#include <linux/bitfield.h>
 #include <linux/firmware.h>
 #include <linux/i2c.h>
 #include <linux/init.h>
 #include <linux/module.h>
-#include <linux/of.h>
 #include <linux/slab.h>
 #include <linux/spinlock.h>
 #include <linux/sysfs.h>
 
+struct eeprom_chip {
+	u16 address_mask;
+	u8 num_address_bytes;
+	bool read_only;
+};
+
 struct eeprom_data {
+	const struct eeprom_chip *chip;
 	struct bin_attribute bin;
 	spinlock_t buffer_lock;
 	u16 buffer_idx;
-	u16 address_mask;
-	u8 num_address_bytes;
 	u8 idx_write_cnt;
-	bool read_only;
 	u8 buffer[];
 };
 
-#define I2C_SLAVE_BYTELEN GENMASK(15, 0)
-#define I2C_SLAVE_FLAG_ADDR16 BIT(16)
-#define I2C_SLAVE_FLAG_RO BIT(17)
-#define I2C_SLAVE_DEVICE_MAGIC(_len, _flags) ((_flags) | ((_len) - 1))
-
 static int i2c_slave_eeprom_slave_cb(struct i2c_client *client,
 				     enum i2c_slave_event event, u8 *val)
 {
@@ -49,17 +46,15 @@ static int i2c_slave_eeprom_slave_cb(struct i2c_client *client,
 
 	switch (event) {
 	case I2C_SLAVE_WRITE_RECEIVED:
-		if (eeprom->idx_write_cnt < eeprom->num_address_bytes) {
+		if (eeprom->idx_write_cnt < eeprom->chip->num_address_bytes) {
 			if (eeprom->idx_write_cnt == 0)
 				eeprom->buffer_idx = 0;
 			eeprom->buffer_idx = *val | (eeprom->buffer_idx << 8);
 			eeprom->idx_write_cnt++;
-		} else {
-			if (!eeprom->read_only) {
-				spin_lock(&eeprom->buffer_lock);
-				eeprom->buffer[eeprom->buffer_idx++ & eeprom->address_mask] = *val;
-				spin_unlock(&eeprom->buffer_lock);
-			}
+		} else if (!eeprom->chip->read_only) {
+			spin_lock(&eeprom->buffer_lock);
+			eeprom->buffer[eeprom->buffer_idx++ & eeprom->chip->address_mask] = *val;
+			spin_unlock(&eeprom->buffer_lock);
 		}
 		break;
 
@@ -69,7 +64,7 @@ static int i2c_slave_eeprom_slave_cb(struct i2c_client *client,
 		fallthrough;
 	case I2C_SLAVE_READ_REQUESTED:
 		spin_lock(&eeprom->buffer_lock);
-		*val = eeprom->buffer[eeprom->buffer_idx & eeprom->address_mask];
+		*val = eeprom->buffer[eeprom->buffer_idx & eeprom->chip->address_mask];
 		spin_unlock(&eeprom->buffer_lock);
 		/*
 		 * Do not increment buffer_idx here, because we don't know if
@@ -142,19 +137,25 @@ static int i2c_slave_init_eeprom_data(struct eeprom_data *eeprom, struct i2c_cli
 
 static int i2c_slave_eeprom_probe(struct i2c_client *client)
 {
-	const struct i2c_device_id *id = i2c_client_get_device_id(client);
+	const struct i2c_device_id *id;
+	const struct eeprom_chip *chip;
 	struct eeprom_data *eeprom;
 	int ret;
-	unsigned int size = FIELD_GET(I2C_SLAVE_BYTELEN, id->driver_data) + 1;
-	unsigned int flag_addr16 = FIELD_GET(I2C_SLAVE_FLAG_ADDR16, id->driver_data);
+	unsigned int size;
+
+	id = i2c_client_get_device_id(client);
+	if (!id)
+		return -EINVAL;
+
+	chip = id->data;
+	size = chip->address_mask + 1;
 
-	eeprom = devm_kzalloc(&client->dev, sizeof(struct eeprom_data) + size, GFP_KERNEL);
+	eeprom = devm_kzalloc(&client->dev, struct_size(eeprom, buffer, size),
+			      GFP_KERNEL);
 	if (!eeprom)
 		return -ENOMEM;
 
-	eeprom->num_address_bytes = flag_addr16 ? 2 : 1;
-	eeprom->address_mask = size - 1;
-	eeprom->read_only = FIELD_GET(I2C_SLAVE_FLAG_RO, id->driver_data);
+	eeprom->chip = chip;
 	spin_lock_init(&eeprom->buffer_lock);
 	i2c_set_clientdata(client, eeprom);
 
@@ -190,15 +191,47 @@ static void i2c_slave_eeprom_remove(struct i2c_client *client)
 	sysfs_remove_bin_file(&client->dev.kobj, &eeprom->bin);
 }
 
+static const struct eeprom_chip chip_24c02 = {
+	.address_mask = 2048 / 8 - 1, .num_address_bytes = 1,
+};
+
+static const struct eeprom_chip chip_24c02ro = {
+	.address_mask = 2048 / 8 - 1, .num_address_bytes = 1, .read_only = true,
+};
+
+static const struct eeprom_chip chip_24c32 = {
+	.address_mask = 32768 / 8 - 1, .num_address_bytes = 2,
+};
+
+static const struct eeprom_chip chip_24c32ro = {
+	.address_mask = 32768 / 8 - 1, .num_address_bytes = 2, .read_only = true,
+};
+
+static const struct eeprom_chip chip_24c64 = {
+	.address_mask = 65536 / 8 - 1, .num_address_bytes = 2,
+};
+
+static const struct eeprom_chip chip_24c64ro = {
+	.address_mask = 65536 / 8 - 1, .num_address_bytes = 2, .read_only = true,
+};
+
+static const struct eeprom_chip chip_24c512 = {
+	.address_mask = 524288 / 8 - 1, .num_address_bytes = 2,
+};
+
+static const struct eeprom_chip chip_24c512ro = {
+	.address_mask = 524288 / 8 - 1, .num_address_bytes = 2, .read_only = true,
+};
+
 static const struct i2c_device_id i2c_slave_eeprom_id[] = {
-	{ "slave-24c02", I2C_SLAVE_DEVICE_MAGIC(2048 / 8,  0) },
-	{ "slave-24c02ro", I2C_SLAVE_DEVICE_MAGIC(2048 / 8,  I2C_SLAVE_FLAG_RO) },
-	{ "slave-24c32", I2C_SLAVE_DEVICE_MAGIC(32768 / 8, I2C_SLAVE_FLAG_ADDR16) },
-	{ "slave-24c32ro", I2C_SLAVE_DEVICE_MAGIC(32768 / 8, I2C_SLAVE_FLAG_ADDR16 | I2C_SLAVE_FLAG_RO) },
-	{ "slave-24c64", I2C_SLAVE_DEVICE_MAGIC(65536 / 8, I2C_SLAVE_FLAG_ADDR16) },
-	{ "slave-24c64ro", I2C_SLAVE_DEVICE_MAGIC(65536 / 8, I2C_SLAVE_FLAG_ADDR16 | I2C_SLAVE_FLAG_RO) },
-	{ "slave-24c512", I2C_SLAVE_DEVICE_MAGIC(524288 / 8, I2C_SLAVE_FLAG_ADDR16) },
-	{ "slave-24c512ro", I2C_SLAVE_DEVICE_MAGIC(524288 / 8, I2C_SLAVE_FLAG_ADDR16 | I2C_SLAVE_FLAG_RO) },
+	{ "slave-24c02",	.data = &chip_24c02 },
+	{ "slave-24c02ro",	.data = &chip_24c02ro },
+	{ "slave-24c32",	.data = &chip_24c32 },
+	{ "slave-24c32ro",	.data = &chip_24c32ro },
+	{ "slave-24c64",	.data = &chip_24c64 },
+	{ "slave-24c64ro",	.data = &chip_24c64ro },
+	{ "slave-24c512",	.data = &chip_24c512 },
+	{ "slave-24c512ro",	.data = &chip_24c512ro },
 	{ }
 };
 MODULE_DEVICE_TABLE(i2c, i2c_slave_eeprom_id);

-- 
2.41.0.694.ge786442a9b-goog

Re: [PATCH RFC 2/4] i2c: mux: ltc4306: convert to using a pointer in i2c_device_id

[Andy Shevchenko]

On Mon, Aug 14, 2023 at 02:52:50PM -0700, Dmitry Torokhov wrote:
> Switch the driver to use newly added "data" pointer in i2c_device_id to
> streamline DT and legacy flows.

...

>  	chip = of_device_get_match_data(&client->dev);
> -
>  	if (!chip)
> -		chip = &chips[i2c_match_id(ltc4306_id, client)->driver_data];
> +		chip = i2c_match_id(ltc4306_id, client)->data;

The whole thing can be converted to i2c_device_get_match_data() with the new
helper from another thread. That said, I don't see value in this half-baked
change.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH RFC 3/4] i2c: mux: pca954x: convert to using a pointer in i2c_device_id

[Andy Shevchenko]

On Mon, Aug 14, 2023 at 02:52:51PM -0700, Dmitry Torokhov wrote:
> Switch the driver to use newly added "data" pointer in i2c_device_id to
> streamline DT and legacy flows.

...

> @@ -206,7 +206,7 @@ static const struct of_device_id pca954x_of_match[] = {
>  	{ .compatible = "nxp,pca9847", .data = &chips[pca_9847] },
>  	{ .compatible = "nxp,pca9848", .data = &chips[pca_9848] },
>  	{ .compatible = "nxp,pca9849", .data = &chips[pca_9849] },
> -	{}
> +	{ }
>  };
>  MODULE_DEVICE_TABLE(of, pca954x_of_match);

Stray change.

...

>  	data->chip = device_get_match_data(dev);
>  	if (!data->chip)
> -		data->chip = &chips[id->driver_data];
> +		data->chip = id->data;

Same, why not to use that helper and if anything, modify it instead.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH RFC 4/4] i2c: slave-eeprom: convert to using a pointer in i2c_device_id

[Andy Shevchenko]

On Mon, Aug 14, 2023 at 02:52:52PM -0700, Dmitry Torokhov wrote:
> Switch the driver to use newly added "data" pointer in i2c_device_id and
> introduce struct eeprom_chip to describe chip's characteristics instead
> of cramming it all into an unisigned long and then decipher.

...

> -	const struct i2c_device_id *id = i2c_client_get_device_id(client);
> +	const struct i2c_device_id *id;

> +	id = i2c_client_get_device_id(client);
> +	if (!id)
> +		return -EINVAL;

Unneeded piece of change.


-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH RFC 1/4] i2c: start migrating to a pointer in i2c_device_id

[Andy Shevchenko]

On Mon, Aug 14, 2023 at 02:52:49PM -0700, Dmitry Torokhov wrote:
> The of_device_id structure uses a void pointer to associate additional
> driver-private data with device id, most commonly used to refer to a
> structure containing additional characteristics of a particular chip.
> However i2c_device_id uses an unsigned long. While it can easily be
> converted to a pointer, several drivers use it to store a scalar (and it
> is possible to use a pointer in of_device_id to store a scalar value as
> well). The worst case comes when OF part of a driver uses pointers,
> while legacy part is using scalars, causing constructs like:
> 
> 	data = device_get_match_data(...);
> 	if (!data)
> 		data = &some_array[i2c_match_id(...)->data];
> 	...
> 
> To avoid this introduce a const void "data" pointer to i2c_device_id as
> well, so that we can convert drivers one by one, and drop driver_data
> member in the end.
> 
> The end goal is to clean up all helpers and make device_get_match_data()
> work universally for all ACPI, DT, and legacy variants.

So, we have in the parallel the activity to make buses to have a callback,
why do we need this one? Looks to me as yet another 1000+ churn for not
much value. What the good outcome of this is constification, but maybe
we can find the way on how to prove const to stay over the kernel_ulong_t
transformations for all device ID tables?

-- 
With Best Regards,
Andy Shevchenko