Skip to contentSkip to conversation

mfd: da9030 usb charge pump support within mfd driver.

PATCH In review Jonathan Cameron 5 participants 14 messages Jan 14, 2009 – Jan 15, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mfd: da9030 usb charge pump support within mfd driver.

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

---

Minor changes from previous version to reflect changes in the driver being patched. Addtional cc's as per Mark's suggestion.

This version simply adds the functionality to the da903x mfd driver core. If anyone can suggest a way round simply maintaining a global pointer to the device it would be good.

To be able to call this from a board config file the driver must be built in rather than a module but enforcing that is down to the board Kconfig entry.

drivers/mfd/da903x.c +29 -0
index 99f8dcf..70ec3b0 100644
@@ -30,6 +30,7 @@
 #define DA9030_SYS_CTRL_A	0x08
 #define DA9030_SYS_CTRL_B	0x09
 #define DA9030_FAULT_LOG	0x0a
+#define DA9030_USBPUMP_REG	0x19
 
 #define DA9034_CHIP_ID		0x00
 #define DA9034_EVENT_A		0x01
@@ -484,6 +485,32 @@ failed:
 	return ret;
 }
 
+/* Single device assumption */
+struct device *da9030_mfd_dev;
+
+int da9030_set_usb_charge_pump_mode(int mode)
+{
+	uint8_t val = 0;
+
+	if (da9030_mfd_dev == NULL)
+		return -EINVAL;
+
+	switch (mode) {
+	case DA9030_USBPUMP_AUTO:
+		val = 0;
+		break;
+	case DA9030_USBPUMP_CHARGE_PUMP:
+		val = 0x40;
+		break;
+	case DA9030_USBPUMP_CURRENT_SOURCE:
+		val = 0x80;
+		break;
+	}
+
+	return da903x_write(da9030_mfd_dev, DA9030_USBPUMP_REG, val);
+}
+EXPORT_SYMBOL_GPL(da9030_set_usb_charge_pump_mode);
+
 static int __devinit da903x_probe(struct i2c_client *client,
 				  const struct i2c_device_id *id)
 {
@@ -499,6 +526,7 @@ static int __devinit da903x_probe(struct i2c_client *client,
 	chip->client = client;
 	chip->dev = &client->dev;
 	chip->ops = &da903x_ops[id->driver_data];
+	da9030_mfd_dev = &client->dev;
 
 	mutex_init(&chip->lock);
 	INIT_WORK(&chip->irq_work, da903x_irq_work);
@@ -559,6 +587,7 @@ static struct i2c_driver da903x_driver = {
 
 static int __init da903x_init(void)
 {
+	da9030_mfd_dev = NULL;
 	return i2c_add_driver(&da903x_driver);
 }
 module_init(da903x_init);
include/linux/mfd/da903x.h +5 -0
index 115dbe9..5e15305 100644
@@ -227,6 +227,11 @@ extern int da903x_unregister_notifier(struct device *dev,
 
 extern int da903x_query_status(struct device *dev, unsigned int status);
 
+#define DA9030_USBPUMP_AUTO		1
+#define DA9030_USBPUMP_CHARGE_PUMP	2
+#define DA9030_USBPUMP_CURRENT_SOURCE	3
+
+int da9030_set_usb_charge_pump_mode(int mode);
 
 /* NOTE: the functions below are not intended for use outside
  * of the DA903x sub-device drivers
#1 lore raw
Quoted 5 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

#2 lore raw
Quoted 9 lines · Eric Miao
Quoted 4 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

It's not exactly related to the charger, it's rather related to the USB voltage supplied to USB devices attached to PXA OHCI.
Indeed the mfd/da903x.c serves as a common core for sub-peripherals, but IMHO adding a subdevice driver because of single method doesn't worth the overhead. I'm for the solution Jonathan proposes.

Signature · 2 lines
Sincerely yours,
Mike.
#3 lore raw
Quoted 13 lines · Mike Rapoport
Quoted 9 lines · Eric Miao
Quoted 4 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

Quoted 6 lines

It's not exactly related to the charger, it's rather related to the USB voltage supplied to USB devices attached to PXA OHCI.
Indeed the mfd/da903x.c serves as a common core for sub-peripherals, but IMHO adding a subdevice driver because of single method doesn't worth the overhead. I'm for the solution Jonathan proposes.

Mmm... then who will be the invoker? I'm a bit upset about this being exported while called out of control. If it works as the name suggested, setting some mode, maybe we can have a platform data field specifying this, and hide this totally within the driver. I didn't look into this too much, just a concern here.

#4 lore raw
Quoted 18 lines · Eric Miao
Quoted 7 lines · Mike Rapoport
Quoted 5 lines · Eric Miao
Quoted 4 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Quoted 9 lines
Quoted 3 lines

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

It's not exactly related to the charger, it's rather related to the USB voltage supplied to USB devices attached to PXA OHCI.
Indeed the mfd/da903x.c serves as a common core for sub-peripherals, but IMHO adding a subdevice driver because of single method doesn't worth the overhead. I'm for the solution Jonathan proposes.

Quoted 5 lines

Mmm... then who will be the invoker? I'm a bit upset about this being exported while called out of control. If it works as the name suggested, setting some mode, maybe we can have a platform data field specifying this, and hide this totally within the driver. I didn't look into this too much, just a concern here.

The USB charge pump should be enabled if DA9030 should supply USB VBUS for USB devices connected to PXA OHCI port. In the most simple case, when the USB port is configured to be host only the platform data can be used to setup the charge pump once and forget about it. However, if the USB port is configured as OTG, the charge pump should be toggled depending on OTG ID.
In EM-X270 I have a GPIO connected to OTG ID pin of the connector and according to the GPIO state I toggle the da9030 USB charge pump.

Signature · 2 lines
Sincerely yours,
Mike.
#5 lore raw
Quoted 27 lines · Mike Rapoport
Quoted 18 lines · Eric Miao
Quoted 7 lines · Mike Rapoport
Quoted 5 lines · Eric Miao
Quoted 4 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Quoted 9 lines
Quoted 3 lines

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

It's not exactly related to the charger, it's rather related to the USB voltage supplied to USB devices attached to PXA OHCI.
Indeed the mfd/da903x.c serves as a common core for sub-peripherals, but IMHO adding a subdevice driver because of single method doesn't worth the overhead. I'm for the solution Jonathan proposes.

Quoted 5 lines

Mmm... then who will be the invoker? I'm a bit upset about this being exported while called out of control. If it works as the name suggested, setting some mode, maybe we can have a platform data field specifying this, and hide this totally within the driver. I didn't look into this too much, just a concern here.

Quoted 8 lines

The USB charge pump should be enabled if DA9030 should supply USB VBUS for USB devices connected to PXA OHCI port. In the most simple case, when the USB port is configured to be host only the platform data can be used to setup the charge pump once and forget about it. However, if the USB port is configured as OTG, the charge pump should be toggled depending on OTG ID.
In EM-X270 I have a GPIO connected to OTG ID pin of the connector and according to the GPIO state I toggle the da9030 USB charge pump.

OK, now I understand the problem. The optimal way would be to invent an abstraction for this external charge pump (in terms of PXA27x OTG controller) and make DA9030 a derived class of it. But the PXA27x OTG solution is really a hybrid one, and the existing OTG code is in a mess (without a clear abstraction), the sub-optimal way would let platform code do this trick. Yet the platform code is currently suffering from a missed API to retreive this dynamically scanned I2C device. Mmm...stumped.

I would expect the final solution to be clean enough, that requires some dependent stuffs to be modified, and the final look of this patch may be a bit different.

#6 lore raw
Quoted 42 lines · Eric Miao
Quoted 30 lines · Mike Rapoport
Quoted 22 lines · Eric Miao
Quoted 13 lines · Mike Rapoport
Quoted 11 lines · Eric Miao
Quoted 5 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Well, if it is totally USB charger related, I'd suggest to move this into the dedicated driver. This mfd/da903x.c serves as a common code base for all sub-peripherals.

Quoted 7 lines

It's not exactly related to the charger, it's rather related to the USB voltage supplied to USB devices attached to PXA OHCI.
Indeed the mfd/da903x.c serves as a common core for sub-peripherals, but IMHO adding a subdevice driver because of single method doesn't worth the overhead. I'm for the solution Jonathan proposes.

Quoted 6 lines

Mmm... then who will be the invoker? I'm a bit upset about this being exported while called out of control. If it works as the name suggested, setting some mode, maybe we can have a platform data field specifying this, and hide this totally within the driver. I didn't look into this too much, just a concern here.

Quoted 9 lines

The USB charge pump should be enabled if DA9030 should supply USB VBUS for USB devices connected to PXA OHCI port. In the most simple case, when the USB port is configured to be host only the platform data can be used to setup the charge pump once and forget about it. However, if the USB port is configured as OTG, the charge pump should be toggled depending on OTG ID.
In EM-X270 I have a GPIO connected to OTG ID pin of the connector and according to the GPIO state I toggle the da9030 USB charge pump.

Quoted 8 lines

OK, now I understand the problem. The optimal way would be to invent an abstraction for this external charge pump (in terms of PXA27x OTG controller) and make DA9030 a derived class of it. But the PXA27x OTG solution is really a hybrid one, and the existing OTG code is in a mess (without a clear abstraction), the sub-optimal way would let platform code do this trick. Yet the platform code is currently suffering from a missed API to retreive this dynamically scanned I2C device. Mmm...stumped.

Agreed, that lack of a means to obtain i2c devices in board configs is beginning to get in the way
of quite a few things. Perhaps we need to start a proper discussion on how to sort that out.
(some sort of registered call backs perhaps to allow the right dev pointers to be set on the i2c
device finally being allocated?) I'm guessing the i2c guys are going to have strong views
on any changes to their allocation approach!

Quoted 4 lines

I would expect the final solution to be clean enough, that requires some dependent stuffs to be modified, and the final look of this patch may be a bit different

Agreed. This definitely isn't the way to go in the long run (assuming otg etc get cleaned up)
but would it be an acceptable stop gap? I'm just keen to get the functionality available
so as to get a board config (intel stargate 2) sorted. (some of which you were kind enough to
review a while back - thanks).

Jonathan

#7 lore raw

")
Fcc: +sent-mail

Quoted 11 lines · Jonathan Cameron
Quoted 3 lines

I would expect the final solution to be clean enough, that requires some dependent stuffs to be modified, and the final look of this patch may be a bit different

Agreed. This definitely isn't the way to go in the long run (assuming otg etc get cleaned up)
but would it be an acceptable stop gap? I'm just keen to get the functionality available
so as to get a board config (intel stargate 2) sorted. (some of which you were kind enough to
review a while back - thanks).

Having looked at this problem just this week for some other designs I'm thinking the regulator API might be a good fit for this. It is a supply and the API provides a method to match up the supply with the USB controller (some designs have multiple options there so that's useful). I've not actually tried it yet to see what the pain is like, though.

#8 lore raw
Quoted 17 lines · Mark Brown

")
Fcc: +sent-mail

Quoted 11 lines · Jonathan Cameron
Quoted 3 lines

I would expect the final solution to be clean enough, that requires some dependent stuffs to be modified, and the final look of this patch may be a bit different

Agreed. This definitely isn't the way to go in the long run (assuming otg etc get cleaned up)
but would it be an acceptable stop gap? I'm just keen to get the functionality available
so as to get a board config (intel stargate 2) sorted. (some of which you were kind enough to
review a while back - thanks).

Quoted 5 lines

Having looked at this problem just this week for some other designs I'm thinking the regulator API might be a good fit for this. It is a supply and the API provides a method to match up the supply with the USB controller (some designs have multiple options there so that's useful). I've not actually tried it yet to see what the pain is like, though.

I did at one point. The problem here is you aren't simply controlling the current / voltage. You are switching it between various automatic modes and a manual override. Only in the manually overridden states is it much like a regulator and then it's one with very odd properties.

If these are consistent enough across different pmic's I guess it could be blugeoned in, but I'm not convinced this is true.

Jonathan

#9 lore raw
Quoted 8 lines · Jonathan Cameron
Quoted 5 lines · Mark Brown

Having looked at this problem just this week for some other designs I'm thinking the regulator API might be a good fit for this. It is a supply and the API provides a method to match up the supply with the USB controller (some designs have multiple options there so that's useful). I've not actually tried it yet to see what the pain is like, though.

Quoted 4 lines

I did at one point. The problem here is you aren't simply controlling the current / voltage. You are switching it between various automatic modes and a manual override. Only in the manually overridden states is it much like a regulator and then it's one with very odd properties.

Yes, they're rather different and would require additional APIs for controlling some of the features, though it may be that we can get away with doing a lot of the configuration in platform data or board init code with a more limited subset configured dynamically by the USB stack.

Quoted 2 lines

If these are consistent enough across different pmic's I guess it could be blugeoned in, but I'm not convinced this is true.

Yes, that needs to be investigated. I'd hope there's *some* similarity since at the end of the day it's all USB but I've not got to the point of looking at that yet.

#10 lore raw
In reply to Jonathan Cameron

Jonathan,
Below are my two cents :)

Quoted 6 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Quoted 8 lines

---

Minor changes from previous version to reflect changes in the driver being patched. Addtional cc's as per Mark's suggestion.

This version simply adds the functionality to the da903x mfd driver core. If anyone can suggest a way round simply maintaining a global pointer to the device it would be good.

I don't think the global pointer here is real problem. It should be *really* weird design with two da903x.

Quoted 3 lines

To be able to call this from a board config file the driver must be built in rather than a module but enforcing that is down to the board Kconfig entry.

Most probably the board with da9030 will anyway have it built-in...

Quoted 32 lines
drivers/mfd/da903x.c +17 -0
index 99f8dcf..70ec3b0 100644
@@ -30,6 +30,7 @@
 #define DA9030_SYS_CTRL_A	0x08
 #define DA9030_SYS_CTRL_B	0x09
 #define DA9030_FAULT_LOG	0x0a
+#define DA9030_USBPUMP_REG	0x19
 
 #define DA9034_CHIP_ID		0x00
 #define DA9034_EVENT_A		0x01
@@ -484,6 +485,32 @@ failed:
 	return ret;
 }
 
+/* Single device assumption */
+struct device *da9030_mfd_dev;
+
+int da9030_set_usb_charge_pump_mode(int mode)
+{
+	uint8_t val = 0;
+
+	if (da9030_mfd_dev == NULL)
+		return -EINVAL;
+
+	switch (mode) {
+	case DA9030_USBPUMP_AUTO:
+		val = 0;
+		break;
+	case DA9030_USBPUMP_CHARGE_PUMP:
+		val = 0x40;

drivers/mfd/da903x.c failed: · line 513

Quoted 10 lines
drivers/mfd/da903x.c +10 -0
+		break;
+	case DA9030_USBPUMP_CURRENT_SOURCE:
+		val = 0x80;
+		break;
+	}
+
+	return da903x_write(da9030_mfd_dev, DA9030_USBPUMP_REG, val);
+}
+EXPORT_SYMBOL_GPL(da9030_set_usb_charge_pump_mode);
+

Maybe we want to have 'da9030_set_usb_charge_pump_mode' as well?

Quoted 41 lines
drivers/mfd/da903x.c +2 -0
 static int __devinit da903x_probe(struct i2c_client *client,
 				  const struct i2c_device_id *id)
 {
@@ -499,6 +526,7 @@ static int __devinit da903x_probe(struct i2c_client *client,
 	chip->client = client;
 	chip->dev = &client->dev;
 	chip->ops = &da903x_ops[id->driver_data];
+	da9030_mfd_dev = &client->dev;
 
 	mutex_init(&chip->lock);
 	INIT_WORK(&chip->irq_work, da903x_irq_work);
@@ -559,6 +587,7 @@ static struct i2c_driver da903x_driver = {
 
 static int __init da903x_init(void)
 {
+	da9030_mfd_dev = NULL;
 	return i2c_add_driver(&da903x_driver);
 }
 module_init(da903x_init);
include/linux/mfd/da903x.h +5 -1
index 115dbe9..5e15305 100644
@@ -227,6 +227,11 @@ extern int da903x_unregister_notifier(struct device *dev,
 
 extern int da903x_query_status(struct device *dev, unsigned int status);
 
+#define DA9030_USBPUMP_AUTO		1
+#define DA9030_USBPUMP_CHARGE_PUMP	2
+#define DA9030_USBPUMP_CURRENT_SOURCE	3
+
+int da9030_set_usb_charge_pump_mode(int mode);
 
 /* NOTE: the functions below are not intended for use outside
  * of the DA903x sub-device drivers
--

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 2 lines
Sincerely yours,
Mike.
#11 lore raw
Quoted 3 lines · Mike Rapoport

Jonathan,
Below are my two cents :)

Thanks,

Quoted 67 lines
Quoted 6 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

Quoted 9 lines

---

Minor changes from previous version to reflect changes in the driver being patched. Addtional cc's as per Mark's suggestion.

This version simply adds the functionality to the da903x mfd driver core. If anyone can suggest a way round simply maintaining a global pointer to the device it would be good.

I don't think the global pointer here is real problem. It should be *really* weird design with two da903x.

Quoted 4 lines

To be able to call this from a board config file the driver must be built in rather than a module but enforcing that is down to the board Kconfig entry.

Most probably the board with da9030 will anyway have it built-in...

Quoted 37 lines
drivers/mfd/da903x.c +22 -0
index 99f8dcf..70ec3b0 100644
@@ -30,6 +30,7 @@
 #define DA9030_SYS_CTRL_A	0x08
 #define DA9030_SYS_CTRL_B	0x09
 #define DA9030_FAULT_LOG	0x0a
+#define DA9030_USBPUMP_REG	0x19
 
 #define DA9034_CHIP_ID		0x00
 #define DA9034_EVENT_A		0x01
@@ -484,6 +485,32 @@ failed:
 	return ret;
 }
 
+/* Single device assumption */
+struct device *da9030_mfd_dev;
+
+int da9030_set_usb_charge_pump_mode(int mode)
+{
+	uint8_t val = 0;
+
+	if (da9030_mfd_dev == NULL)
+		return -EINVAL;
+
+	switch (mode) {
+	case DA9030_USBPUMP_AUTO:
+		val = 0;
+		break;
+	case DA9030_USBPUMP_CHARGE_PUMP:
+		val = 0x40;
+		break;
+	case DA9030_USBPUMP_CURRENT_SOURCE:
+		val = 0x80;
+		break;
+	}
Quoted 9 lines

drivers/mfd/da903x.c failed:

Quoted 6 lines
drivers/mfd/da903x.c +5 -0
+
+	return da903x_write(da9030_mfd_dev, DA9030_USBPUMP_REG, val);
+}
+EXPORT_SYMBOL_GPL(da9030_set_usb_charge_pump_mode);
+
 

Maybe we want to have 'da9030_set_usb_charge_pump_mode' as well?

(get?)
Possibly though not sure under what circumstances it would be needed other than debugging?

#12 lore raw
Quoted 10 lines · Jonathan Cameron
Quoted 3 lines · Mike Rapoport

Jonathan,
Below are my two cents :)

Thanks,

Quoted 3 lines
Quoted 1 line

Maybe we want to have 'da9030_set_usb_charge_pump_mode' as well?

(get?)

get of course :)

Quoted 3 lines

Possibly though not sure under what circumstances it would be needed other than debugging?

Somehow I thought I used it in the battery charger, but it seems I was wrong.

Signature · 2 lines
Sincerely yours,
Mike.
#13 lore raw
In reply to Mike Rapoport
Quoted 2 lines · Mike Rapoport

Jonathan,
Below are my two cents :)

Thanks.

Quoted 12 lines
Quoted 11 lines · Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add support for changing the mode of the da9030 usb charge pump

---

Minor changes from previous version to reflect changes in the driver being patched. Addtional cc's as per Mark's suggestion.

Quoted 6 lines
Quoted 3 lines

This version simply adds the functionality to the da903x mfd driver core. If anyone can suggest a way round simply maintaining a global pointer to the device it would be good.

I don't think the global pointer here is real problem. It should be *really* weird design with two da903x.

Would be 'unusual'!

Quoted 5 lines
Quoted 3 lines

To be able to call this from a board config file the driver must be built in rather than a module but enforcing that is down to the board Kconfig entry.

Most probably the board with da9030 will anyway have it built-in...

Agreed.

Quoted 4 lines

drivers/mfd/da903x.c failed: · line 513

Quoted 2 lines
drivers/mfd/da903x.c +2 -0
+EXPORT_SYMBOL_GPL(da9030_set_usb_charge_pump_mode);
+

Maybe we want to have 'da9030_set_usb_charge_pump_mode' as well?

(get?)
Possibly, though only purpose I can think of for this would be debugging and from the datasheet I can't see any case under which the set can fail?