Skip to contentSkip to conversation

gpio: sim: don't fiddle with GPIOLIB private members

PATCH Superseded Bartosz Golaszewski 2 participants 5 messages Sep 7, 2023 – Sep 11, 2023 GPIO

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH v4 1/1 gpio: sim: don't fiddle with GPIOLIB private members

From: Bartosz Golaszewski <[email protected]>

We access internals of struct gpio_device and struct gpio_desc because it's easier but it can actually be avoided and we're working towards a better encapsulation of GPIO data structures across the kernel so let's start at home.

Instead of checking gpio_desc flags, let's just track the requests of GPIOs in the driver. We also already store the information about direction of simulated lines.

For kobjects needed by sysfs callbacks: we can iterate over the children devices of the top-level platform device and compare their fwnodes against the one passed to the init function from probe.

While at it: fix one line break and remove the untrue part about configfs callbacks using dev_get_drvdata() from a comment.

---
v1 -> v2:
- use get_dev_from_fwnode() instead of dereferencing fwnode directly

v2 -> v3:
- don't use fwnode internal fields, instead: iterate over the platform device's children and locate the GPIO device

v3 -> v4:
- use device_find_child() (not bus_device_find_by_fwnode() as it's pointless to iterate over all platform devices)
- use device_match_fwnode() for matching the device by fwnode

 drivers/gpio/gpio-sim.c | 75 ++++++++++++++++++++++++++++++-----------
 1 file changed, 56 insertions(+), 19 deletions(-)
drivers/gpio/gpio-sim.c +56 -19
index 271db3639a78..a58387f2cdc6 100644
@@ -12,6 +12,8 @@
 #include <linux/completion.h>
 #include <linux/configfs.h>
 #include <linux/device.h>
+#include <linux/device/bus.h>
+#include <linux/gpio/consumer.h>
 #include <linux/gpio/driver.h>
 #include <linux/gpio/machine.h>
 #include <linux/idr.h>
@@ -30,8 +32,6 @@
 #include <linux/string_helpers.h>
 #include <linux/sysfs.h>
 
-#include "gpiolib.h"
-
 #define GPIO_SIM_NGPIO_MAX	1024
 #define GPIO_SIM_PROP_MAX	4 /* Max 3 properties + sentinel. */
 #define GPIO_SIM_NUM_ATTRS	3 /* value, pull and sentinel */
@@ -40,6 +40,8 @@ static DEFINE_IDA(gpio_sim_ida);
 
 struct gpio_sim_chip {
 	struct gpio_chip gc;
+	struct device *dev;
+	unsigned long *request_map;
 	unsigned long *direction_map;
 	unsigned long *value_map;
 	unsigned long *pull_map;
@@ -63,16 +65,11 @@ static int gpio_sim_apply_pull(struct gpio_sim_chip *chip,
 			       unsigned int offset, int value)
 {
 	int irq, irq_type, ret;
-	struct gpio_desc *desc;
-	struct gpio_chip *gc;
-
-	gc = &chip->gc;
-	desc = &gc->gpiodev->descs[offset];
 
 	guard(mutex)(&chip->lock);
 
-	if (test_bit(FLAG_REQUESTED, &desc->flags) &&
-	    !test_bit(FLAG_IS_OUT, &desc->flags)) {
+	if (test_bit(offset, chip->request_map) &&
+	    test_bit(offset, chip->direction_map)) {
 		if (value == !!test_bit(offset, chip->value_map))
 			goto set_pull;
 
@@ -99,8 +96,8 @@ static int gpio_sim_apply_pull(struct gpio_sim_chip *chip,
 
 set_value:
 	/* Change the value unless we're actively driving the line. */
-	if (!test_bit(FLAG_REQUESTED, &desc->flags) ||
-	    !test_bit(FLAG_IS_OUT, &desc->flags))
+	if (!test_bit(offset, chip->request_map) ||
+	    test_bit(offset, chip->direction_map))
 		__assign_bit(offset, chip->value_map, value);
 
 set_pull:
@@ -180,8 +177,8 @@ static int gpio_sim_get_direction(struct gpio_chip *gc, unsigned int offset)
 	return direction ? GPIO_LINE_DIRECTION_IN : GPIO_LINE_DIRECTION_OUT;
 }
 
-static int gpio_sim_set_config(struct gpio_chip *gc,
-				  unsigned int offset, unsigned long config)
+static int gpio_sim_set_config(struct gpio_chip *gc, unsigned int offset,
+			       unsigned long config)
 {
 	struct gpio_sim_chip *chip = gpiochip_get_data(gc);
 
@@ -204,13 +201,25 @@ static int gpio_sim_to_irq(struct gpio_chip *gc, unsigned int offset)
 	return irq_create_mapping(chip->irq_sim, offset);
 }
 
+static int gpio_sim_request(struct gpio_chip *gc, unsigned int offset)
+{
+	struct gpio_sim_chip *chip = gpiochip_get_data(gc);
+
+	scoped_guard(mutex, &chip->lock)
+		__set_bit(offset, chip->request_map);
+
+	return 0;
+}
+
 static void gpio_sim_free(struct gpio_chip *gc, unsigned int offset)
 {
 	struct gpio_sim_chip *chip = gpiochip_get_data(gc);
 
-	scoped_guard(mutex, &chip->lock)
+	scoped_guard(mutex, &chip->lock) {
 		__assign_bit(offset, chip->value_map,
 			     !!test_bit(offset, chip->pull_map));
+		__clear_bit(offset, chip->request_map);
+	}
 }
 
 static ssize_t gpio_sim_sysfs_val_show(struct device *dev,
@@ -282,6 +291,13 @@ static void gpio_sim_mutex_destroy(void *data)
 	mutex_destroy(lock);
 }
 
+static void gpio_sim_put_device(void *data)
+{
+	struct device *dev = data;
+
+	put_device(dev);
+}
+
 static void gpio_sim_dispose_mappings(void *data)
 {
 	struct gpio_sim_chip *chip = data;
@@ -295,7 +311,7 @@ static void gpio_sim_sysfs_remove(void *data)
 {
 	struct gpio_sim_chip *chip = data;
 
-	sysfs_remove_groups(&chip->gc.gpiodev->dev.kobj, chip->attr_groups);
+	sysfs_remove_groups(&chip->dev->kobj, chip->attr_groups);
 }
 
 static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip)
@@ -352,14 +368,22 @@ static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip)
 		chip->attr_groups[i] = attr_group;
 	}
 
-	ret = sysfs_create_groups(&chip->gc.gpiodev->dev.kobj,
-				  chip->attr_groups);
+	ret = sysfs_create_groups(&chip->dev->kobj, chip->attr_groups);
 	if (ret)
 		return ret;
 
 	return devm_add_action_or_reset(dev, gpio_sim_sysfs_remove, chip);
 }
 
+static int gpio_sim_dev_match_fwnode(struct device *dev, void *data)
+{
+	/*
+	 * We can't pass this directly to device_find_child() due to pointer
+	 * type mismatch.
+	 */
+	return device_match_fwnode(dev, data);
+}
+
 static int gpio_sim_add_bank(struct fwnode_handle *swnode, struct device *dev)
 {
 	struct gpio_sim_chip *chip;
@@ -387,6 +411,10 @@ static int gpio_sim_add_bank(struct fwnode_handle *swnode, struct device *dev)
 	if (!chip)
 		return -ENOMEM;
 
+	chip->request_map = devm_bitmap_zalloc(dev, num_lines, GFP_KERNEL);
+	if (!chip->request_map)
+		return -ENOMEM;
+
 	chip->direction_map = devm_bitmap_alloc(dev, num_lines, GFP_KERNEL);
 	if (!chip->direction_map)
 		return -ENOMEM;
@@ -432,6 +460,7 @@ static int gpio_sim_add_bank(struct fwnode_handle *swnode, struct device *dev)
 	gc->get_direction = gpio_sim_get_direction;
 	gc->set_config = gpio_sim_set_config;
 	gc->to_irq = gpio_sim_to_irq;
+	gc->request = gpio_sim_request;
 	gc->free = gpio_sim_free;
 	gc->can_sleep = true;
 
@@ -439,8 +468,16 @@ static int gpio_sim_add_bank(struct fwnode_handle *swnode, struct device *dev)
 	if (ret)
 		return ret;
 
-	/* Used by sysfs and configfs callbacks. */
-	dev_set_drvdata(&gc->gpiodev->dev, chip);
+	chip->dev = device_find_child(dev, swnode, gpio_sim_dev_match_fwnode);
+	if (!chip->dev)
+		return -ENODEV;
+
+	ret = devm_add_action_or_reset(dev, gpio_sim_put_device, chip->dev);
+	if (ret)
+		return ret;
+
+	/* Used by sysfs callbacks. */
+	dev_set_drvdata(chip->dev, chip);
 
 	return gpio_sim_setup_sysfs(chip);
 }
Signature · 1 line
2.39.2
#1 lore raw
Quoted 11 lines · Bartosz Golaszewski

From: Bartosz Golaszewski <[email protected]>

We access internals of struct gpio_device and struct gpio_desc because it's easier but it can actually be avoided and we're working towards a better encapsulation of GPIO data structures across the kernel so let's start at home.

Instead of checking gpio_desc flags, let's just track the requests of GPIOs in the driver. We also already store the information about direction of simulated lines.

Quoted 6 lines

For kobjects needed by sysfs callbacks: we can iterate over the children devices of the top-level platform device and compare their fwnodes against the one passed to the init function from probe.

While at it: fix one line break and remove the untrue part about configfs callbacks using dev_get_drvdata() from a comment.

Will LGTM with the couple of remarks being addressed.

...

Quoted 5 lines

#include <linux/completion.h>
#include <linux/configfs.h>
#include <linux/device.h>

+#include <linux/device/bus.h>

No need, the device.h guarantees that.

...

drivers/gpio/gpio-sim.c static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip) · line 383

Quoted 6 lines
drivers/gpio/gpio-sim.c +6 -0
+static int gpio_sim_dev_match_fwnode(struct device *dev, void *data)
+{
+	/*
+	 * We can't pass this directly to device_find_child() due to pointer
+	 * type mismatch.
+	 */

Not sure if this comment adds any value.

Quoted 2 lines
drivers/gpio/gpio-sim.c +2 -0
+	return device_match_fwnode(dev, data);
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#2 lore raw
Quoted 24 lines · Andy Shevchenko
Quoted 11 lines · Bartosz Golaszewski

From: Bartosz Golaszewski <[email protected]>

We access internals of struct gpio_device and struct gpio_desc because it's easier but it can actually be avoided and we're working towards a better encapsulation of GPIO data structures across the kernel so let's start at home.

Instead of checking gpio_desc flags, let's just track the requests of GPIOs in the driver. We also already store the information about direction of simulated lines.

Quoted 6 lines

For kobjects needed by sysfs callbacks: we can iterate over the children devices of the top-level platform device and compare their fwnodes against the one passed to the init function from probe.

While at it: fix one line break and remove the untrue part about configfs callbacks using dev_get_drvdata() from a comment.

Will LGTM with the couple of remarks being addressed.

...

Quoted 8 lines
Quoted 5 lines

#include <linux/completion.h>
#include <linux/configfs.h>
#include <linux/device.h>

+#include <linux/device/bus.h>

No need, the device.h guarantees that.

Wait, wasn't you the one who always suggests including headers directly if we're using any symbols defined in them? Like when I said that we don't need to include linux/notifier.h because it's already included in gpiolib.h and you argued the opposite? :)

device_match_fwnode() is defined in linux/device/bus.h so I thought it's in order to include it.

Quoted 11 lines

...

drivers/gpio/gpio-sim.c static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip) · line 383

Quoted 6 lines
drivers/gpio/gpio-sim.c +6 -0
+static int gpio_sim_dev_match_fwnode(struct device *dev, void *data)
+{
+     /*
+      * We can't pass this directly to device_find_child() due to pointer
+      * type mismatch.
+      */

Not sure if this comment adds any value.

I disagree - I would have used device_match_fwnode() as argument passed directly to device_find_child() but I cannot due to pointer type mismatch error so we need this wrapper and it's useful to say why.

Bart

Quoted 8 lines

drivers/gpio/gpio-sim.c static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip) · line 385

Quoted 2 lines
drivers/gpio/gpio-sim.c +2 -0
+     return device_match_fwnode(dev, data);
+}

--
With Best Regards,
Andy Shevchenko

#3 lore raw
Quoted 3 lines · Bartosz Golaszewski
Quoted 1 line · Andy Shevchenko

On Thu, Sep 07, 2023 at 10:27:51AM +0200, Bartosz Golaszewski wrote:

...

Quoted 8 lines
Quoted 7 lines
Quoted 5 lines

#include <linux/completion.h>
#include <linux/configfs.h>
#include <linux/device.h>

+#include <linux/device/bus.h>

No need, the device.h guarantees that.

Quoted 7 lines

Wait, wasn't you the one who always suggests including headers directly if we're using any symbols defined in them? Like when I said that we don't need to include linux/notifier.h because it's already included in gpiolib.h and you argued the opposite? :)

device_match_fwnode() is defined in linux/device/bus.h so I thought it's in order to include it.

Yes, but I am not radical with it, I am for a compromise when some headers guarantee to include some others. That is the case I believe, I don't think device.h ever will be broken to the parts that are not include each other (too many things to change right now, if it happens, not in the feasible
future).

...

Quoted 9 lines
Quoted 8 lines

drivers/gpio/gpio-sim.c static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip) · line 383

Quoted 6 lines
drivers/gpio/gpio-sim.c +6 -0
+static int gpio_sim_dev_match_fwnode(struct device *dev, void *data)
+{
+     /*
+      * We can't pass this directly to device_find_child() due to pointer
+      * type mismatch.
+      */

Not sure if this comment adds any value.

Quoted 4 lines

I disagree - I would have used device_match_fwnode() as argument passed directly to device_find_child() but I cannot due to pointer type mismatch error so we need this wrapper and it's useful to say why.

Yes, and we have dozen(s ?) of the similar wrappers without a comment. So, I'm still for removing it.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
drivers/gpio/gpio-sim.c +2 -0
+     return device_match_fwnode(dev, data);
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#4 lore raw
Quoted 41 lines · Andy Shevchenko
Quoted 3 lines · Bartosz Golaszewski
Quoted 1 line · Andy Shevchenko

On Thu, Sep 07, 2023 at 10:27:51AM +0200, Bartosz Golaszewski wrote:

...

Quoted 8 lines
Quoted 7 lines
Quoted 5 lines

#include <linux/completion.h>
#include <linux/configfs.h>
#include <linux/device.h>

+#include <linux/device/bus.h>

No need, the device.h guarantees that.

Quoted 7 lines

Wait, wasn't you the one who always suggests including headers directly if we're using any symbols defined in them? Like when I said that we don't need to include linux/notifier.h because it's already included in gpiolib.h and you argued the opposite? :)

device_match_fwnode() is defined in linux/device/bus.h so I thought it's in order to include it.

Yes, but I am not radical with it, I am for a compromise when some headers guarantee to include some others. That is the case I believe, I don't think device.h ever will be broken to the parts that are not include each other (too many things to change right now, if it happens, not in the feasible
future).

...

Quoted 9 lines
Quoted 8 lines

drivers/gpio/gpio-sim.c static int gpio_sim_setup_sysfs(struct gpio_sim_chip *chip) · line 383

Quoted 6 lines
drivers/gpio/gpio-sim.c +6 -0
+static int gpio_sim_dev_match_fwnode(struct device *dev, void *data)
+{
+     /*
+      * We can't pass this directly to device_find_child() due to pointer
+      * type mismatch.
+      */

Not sure if this comment adds any value.

Quoted 8 lines
Quoted 4 lines

I disagree - I would have used device_match_fwnode() as argument passed directly to device_find_child() but I cannot due to pointer type mismatch error so we need this wrapper and it's useful to say why.

Yes, and we have dozen(s ?) of the similar wrappers without a comment. So, I'm still for removing it.

Meh, fair enough.

Bart