[RFT][PATCH 1/3] regulator: ab8500-ext: Don't allow set idle mode if info->cfg->hwreq is set

[RFT][PATCH 1/3] regulator: ab8500-ext: Don't allow set idle mode if info->cfg->hwreq is set

[Axel Lin]

The regulator always on with high power mode if info->cfg->hwreq is set.

If we allow set idle mode when info->cfg->hwreq is set, get_mode() returns
REGULATOR_MODE_IDLE but the regulator actually is in REGULATOR_MODE_NORMAL mode.
This patch avoid inconsistent status return by get_mode().

Signed-off-by: Axel Lin <axel.lin@ingics.com>
---
 drivers/regulator/ab8500-ext.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
index 9aee21c..f0a1bbf 100644
--- a/drivers/regulator/ab8500-ext.c
+++ b/drivers/regulator/ab8500-ext.c
@@ -192,6 +192,10 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
 		info->update_val = info->update_val_hp;
 		break;
 	case REGULATOR_MODE_IDLE:
+		/* Always on with high power mode if info->cfg->hwreq is set */
+		if (info->cfg && info->cfg->hwreq)
+			return -EINVAL;
+
 		info->update_val = info->update_val_lp;
 		break;
 
-- 
1.7.10.4

[RFT][PATCH 2/3] regulator: ab8500-ext: Don't update info->update_val if set_mode() fails

[Axel Lin]

This ensures info->update_val status is still correct if set_mode() call fails.
Otherwise, get_mode() may return wrong status if a set_mode() call fails.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
---
 drivers/regulator/ab8500-ext.c |   19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
index f0a1bbf..1efe702 100644
--- a/drivers/regulator/ab8500-ext.c
+++ b/drivers/regulator/ab8500-ext.c
@@ -181,6 +181,7 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
 {
 	int ret = 0;
 	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
+	u8 regval;
 
 	if (info == NULL) {
 		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
@@ -189,14 +190,14 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
 
 	switch (mode) {
 	case REGULATOR_MODE_NORMAL:
-		info->update_val = info->update_val_hp;
+		regval = info->update_val_hp;
 		break;
 	case REGULATOR_MODE_IDLE:
 		/* Always on with high power mode if info->cfg->hwreq is set */
 		if (info->cfg && info->cfg->hwreq)
 			return -EINVAL;
 
-		info->update_val = info->update_val_lp;
+		regval = info->update_val_lp;
 		break;
 
 	default:
@@ -204,12 +205,14 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
 	}
 
 	if (ab8500_ext_regulator_is_enabled(rdev)) {
-		u8 regval;
-
-		ret = enable(info, &regval);
-		if (ret < 0)
+		ret = abx500_mask_and_set_register_interruptible(info->dev,
+					info->update_bank, info->update_reg,
+					info->update_mask, regval);
+		if (ret < 0) {
 			dev_err(rdev_get_dev(rdev),
 				"Could not set regulator mode.\n");
+			return ret;
+		}
 
 		dev_dbg(rdev_get_dev(rdev),
 			"%s-set_mode (bank, reg, mask, value): "
@@ -218,7 +221,9 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
 			info->update_mask, regval);
 	}
 
-	return ret;
+	info->update_val = regval;
+
+	return 0;
 }
 
 static unsigned int ab8500_ext_regulator_get_mode(struct regulator_dev *rdev)
-- 
1.7.10.4

Re: [RFT][PATCH 2/3] regulator: ab8500-ext: Don't update info->update_val if set_mode() fails

[Bengt Jönsson]

On 04/10/2013 02:55 PM, Axel Lin wrote:
> This ensures info->update_val status is still correct if set_mode() call fails.
> Otherwise, get_mode() may return wrong status if a set_mode() call fails.
>
> Signed-off-by: Axel Lin <axel.lin@ingics.com>
> ---
>   drivers/regulator/ab8500-ext.c |   19 ++++++++++++-------
>   1 file changed, 12 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
> index f0a1bbf..1efe702 100644
> --- a/drivers/regulator/ab8500-ext.c
> +++ b/drivers/regulator/ab8500-ext.c
> @@ -181,6 +181,7 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
>   {
>   	int ret = 0;
>   	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
> +	u8 regval;
>   
>   	if (info == NULL) {
>   		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
> @@ -189,14 +190,14 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
>   
>   	switch (mode) {
>   	case REGULATOR_MODE_NORMAL:
> -		info->update_val = info->update_val_hp;
> +		regval = info->update_val_hp;
>   		break;
>   	case REGULATOR_MODE_IDLE:
>   		/* Always on with high power mode if info->cfg->hwreq is set */
>   		if (info->cfg && info->cfg->hwreq)
>   			return -EINVAL;
>   
> -		info->update_val = info->update_val_lp;
> +		regval = info->update_val_lp;
>   		break;
>   
>   	default:
> @@ -204,12 +205,14 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
>   	}
>   
>   	if (ab8500_ext_regulator_is_enabled(rdev)) {
> -		u8 regval;
> -
> -		ret = enable(info, &regval);
> -		if (ret < 0)
> +		ret = abx500_mask_and_set_register_interruptible(info->dev,
> +					info->update_bank, info->update_reg,
> +					info->update_mask, regval);
> +		if (ret < 0) {
>   			dev_err(rdev_get_dev(rdev),
>   				"Could not set regulator mode.\n");
> +			return ret;
> +		}
>   
>   		dev_dbg(rdev_get_dev(rdev),
>   			"%s-set_mode (bank, reg, mask, value): "
> @@ -218,7 +221,9 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
>   			info->update_mask, regval);
>   	}
>   
> -	return ret;
> +	info->update_val = regval;
> +
> +	return 0;
>   }
>   
>   static unsigned int ab8500_ext_regulator_get_mode(struct regulator_dev *rdev)
This patch is needed to fix the error path but it partly depends on the 
previous patch so I would like to get that one sorted out before ack-ing 
this patch.

Regards,

Bengt

[RFT][PATCH 3/3] regulator: ab8500-ext: Remove enable() and disable() functions

[Axel Lin]

Both enable() and disable() functions have only one caller, thus remove them.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
---
 drivers/regulator/ab8500-ext.c |   62 ++++++++++++++--------------------------
 1 file changed, 22 insertions(+), 40 deletions(-)

diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
index 1efe702..dd582be 100644
--- a/drivers/regulator/ab8500-ext.c
+++ b/drivers/regulator/ab8500-ext.c
@@ -54,32 +54,44 @@ struct ab8500_ext_regulator_info {
 	u8 update_val_hw;
 };
 
-static int enable(struct ab8500_ext_regulator_info *info, u8 *regval)
+static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
 {
 	int ret;
+	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
+	u8 regval;
 
-	*regval = info->update_val;
+	if (info == NULL) {
+		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
+		return -EINVAL;
+	}
 
 	/*
 	 * To satisfy both HW high power request and SW request, the regulator
 	 * must be on in high power.
 	 */
 	if (info->cfg && info->cfg->hwreq)
-		*regval = info->update_val_hp;
+		regval = info->update_val_hp;
+	else
+		regval = info->update_val;
 
 	ret = abx500_mask_and_set_register_interruptible(info->dev,
 		info->update_bank, info->update_reg,
-		info->update_mask, *regval);
+		info->update_mask, regval);
 	if (ret < 0) {
 		dev_err(rdev_get_dev(info->rdev),
 			"couldn't set enable bits for regulator\n");
 		return ret;
 	}
 
-	return ret;
+	dev_dbg(rdev_get_dev(rdev), "%s-enable (bank, reg, mask, value):"
+		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
+		info->desc.name, info->update_bank, info->update_reg,
+		info->update_mask, regval);
+
+	return 0;
 }
 
-static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
+static int ab8500_ext_regulator_disable(struct regulator_dev *rdev)
 {
 	int ret;
 	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
@@ -90,53 +102,23 @@ static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
 		return -EINVAL;
 	}
 
-	ret = enable(info, &regval);
-
-	dev_dbg(rdev_get_dev(rdev), "%s-enable (bank, reg, mask, value):"
-		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
-		info->desc.name, info->update_bank, info->update_reg,
-		info->update_mask, regval);
-
-	return ret;
-}
-
-static int disable(struct ab8500_ext_regulator_info *info, u8 *regval)
-{
-	int ret;
-
-	*regval = 0x0;
-
 	/*
 	 * Set the regulator in HW request mode if configured
 	 */
 	if (info->cfg && info->cfg->hwreq)
-		*regval = info->update_val_hw;
+		regval = info->update_val_hw;
+	else
+		regval = 0;
 
 	ret = abx500_mask_and_set_register_interruptible(info->dev,
 		info->update_bank, info->update_reg,
-		info->update_mask, *regval);
+		info->update_mask, regval);
 	if (ret < 0) {
 		dev_err(rdev_get_dev(info->rdev),
 			"couldn't set disable bits for regulator\n");
 		return ret;
 	}
 
-	return ret;
-}
-
-static int ab8500_ext_regulator_disable(struct regulator_dev *rdev)
-{
-	int ret;
-	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
-	u8 regval;
-
-	if (info == NULL) {
-		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
-		return -EINVAL;
-	}
-
-	ret = disable(info, &regval);
-
 	dev_dbg(rdev_get_dev(rdev), "%s-disable (bank, reg, mask, value):"
 		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
 		info->desc.name, info->update_bank, info->update_reg,
-- 
1.7.10.4

Re: [RFT][PATCH 3/3] regulator: ab8500-ext: Remove enable() and disable() functions

[Bengt Jönsson]

On 04/10/2013 02:56 PM, Axel Lin wrote:
> Both enable() and disable() functions have only one caller, thus remove them.
>
> Signed-off-by: Axel Lin <axel.lin@ingics.com>
This patch depends on the first one in the series so I would like to get 
first patch sorted out first.

If the first patch goes in, I am fine with this one too.

Regards,

Bengt
> ---
>   drivers/regulator/ab8500-ext.c |   62 ++++++++++++++--------------------------
>   1 file changed, 22 insertions(+), 40 deletions(-)
>
> diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
> index 1efe702..dd582be 100644
> --- a/drivers/regulator/ab8500-ext.c
> +++ b/drivers/regulator/ab8500-ext.c
> @@ -54,32 +54,44 @@ struct ab8500_ext_regulator_info {
>   	u8 update_val_hw;
>   };
>   
> -static int enable(struct ab8500_ext_regulator_info *info, u8 *regval)
> +static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
>   {
>   	int ret;
> +	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
> +	u8 regval;
>   
> -	*regval = info->update_val;
> +	if (info == NULL) {
> +		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
> +		return -EINVAL;
> +	}
>   
>   	/*
>   	 * To satisfy both HW high power request and SW request, the regulator
>   	 * must be on in high power.
>   	 */
>   	if (info->cfg && info->cfg->hwreq)
> -		*regval = info->update_val_hp;
> +		regval = info->update_val_hp;
> +	else
> +		regval = info->update_val;
>   
>   	ret = abx500_mask_and_set_register_interruptible(info->dev,
>   		info->update_bank, info->update_reg,
> -		info->update_mask, *regval);
> +		info->update_mask, regval);
>   	if (ret < 0) {
>   		dev_err(rdev_get_dev(info->rdev),
>   			"couldn't set enable bits for regulator\n");
>   		return ret;
>   	}
>   
> -	return ret;
> +	dev_dbg(rdev_get_dev(rdev), "%s-enable (bank, reg, mask, value):"
> +		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
> +		info->desc.name, info->update_bank, info->update_reg,
> +		info->update_mask, regval);
> +
> +	return 0;
>   }
>   
> -static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
> +static int ab8500_ext_regulator_disable(struct regulator_dev *rdev)
>   {
>   	int ret;
>   	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
> @@ -90,53 +102,23 @@ static int ab8500_ext_regulator_enable(struct regulator_dev *rdev)
>   		return -EINVAL;
>   	}
>   
> -	ret = enable(info, &regval);
> -
> -	dev_dbg(rdev_get_dev(rdev), "%s-enable (bank, reg, mask, value):"
> -		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
> -		info->desc.name, info->update_bank, info->update_reg,
> -		info->update_mask, regval);
> -
> -	return ret;
> -}
> -
> -static int disable(struct ab8500_ext_regulator_info *info, u8 *regval)
> -{
> -	int ret;
> -
> -	*regval = 0x0;
> -
>   	/*
>   	 * Set the regulator in HW request mode if configured
>   	 */
>   	if (info->cfg && info->cfg->hwreq)
> -		*regval = info->update_val_hw;
> +		regval = info->update_val_hw;
> +	else
> +		regval = 0;
>   
>   	ret = abx500_mask_and_set_register_interruptible(info->dev,
>   		info->update_bank, info->update_reg,
> -		info->update_mask, *regval);
> +		info->update_mask, regval);
>   	if (ret < 0) {
>   		dev_err(rdev_get_dev(info->rdev),
>   			"couldn't set disable bits for regulator\n");
>   		return ret;
>   	}
>   
> -	return ret;
> -}
> -
> -static int ab8500_ext_regulator_disable(struct regulator_dev *rdev)
> -{
> -	int ret;
> -	struct ab8500_ext_regulator_info *info = rdev_get_drvdata(rdev);
> -	u8 regval;
> -
> -	if (info == NULL) {
> -		dev_err(rdev_get_dev(rdev), "regulator info null pointer\n");
> -		return -EINVAL;
> -	}
> -
> -	ret = disable(info, &regval);
> -
>   	dev_dbg(rdev_get_dev(rdev), "%s-disable (bank, reg, mask, value):"
>   		" 0x%02x, 0x%02x, 0x%02x, 0x%02x\n",
>   		info->desc.name, info->update_bank, info->update_reg,

Re: [RFT][PATCH 1/3] regulator: ab8500-ext: Don't allow set idle mode if info->cfg->hwreq is set

[Bengt Jönsson]

On 04/10/2013 02:54 PM, Axel Lin wrote:
> The regulator always on with high power mode if info->cfg->hwreq is set.
>
> If we allow set idle mode when info->cfg->hwreq is set, get_mode() returns
> REGULATOR_MODE_IDLE but the regulator actually is in REGULATOR_MODE_NORMAL mode.
> This patch avoid inconsistent status return by get_mode().
>
> Signed-off-by: Axel Lin <axel.lin@ingics.com>z
> ---
>   drivers/regulator/ab8500-ext.c |    4 ++++
>   1 file changed, 4 insertions(+)
>
> diff --git a/drivers/regulator/ab8500-ext.c b/drivers/regulator/ab8500-ext.c
> index 9aee21c..f0a1bbf 100644
> --- a/drivers/regulator/ab8500-ext.c
> +++ b/drivers/regulator/ab8500-ext.c
> @@ -192,6 +192,10 @@ static int ab8500_ext_regulator_set_mode(struct regulator_dev *rdev,
>   		info->update_val = info->update_val_hp;
>   		break;
>   	case REGULATOR_MODE_IDLE:
> +		/* Always on with high power mode if info->cfg->hwreq is set */
> +		if (info->cfg && info->cfg->hwreq)
> +			return -EINVAL;
> +
>   		info->update_val = info->update_val_lp;
>   		break;
>   
I prefer to have info->update_val updated to reflect requested mode (in 
get_mode) instead of returning -EINVAL. This is how I interpret Mark's 
comment on the other patch ([PATCH] regulator: ab8500: Fix get_mode for 
shared mode regulators).Otherwise, the patch should work fine.

Regards,

Bengt

Re: [RFT][PATCH 1/3] regulator: ab8500-ext: Don't allow set idle mode if info->cfg->hwreq is set

[Axel Lin]

2013/4/16 Bengt Jönsson <bengt.g.jonsson@stericsson.com>:
> On 04/10/2013 02:54 PM, Axel Lin wrote:
>>
>> The regulator always on with high power mode if info->cfg->hwreq is set.
>>
>> If we allow set idle mode when info->cfg->hwreq is set, get_mode() returns
>> REGULATOR_MODE_IDLE but the regulator actually is in REGULATOR_MODE_NORMAL
>> mode.
>> This patch avoid inconsistent status return by get_mode().
>>
>> Signed-off-by: Axel Lin <axel.lin@ingics.com>z
>>
>> ---
>>   drivers/regulator/ab8500-ext.c |    4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/regulator/ab8500-ext.c
>> b/drivers/regulator/ab8500-ext.c
>> index 9aee21c..f0a1bbf 100644
>> --- a/drivers/regulator/ab8500-ext.c
>> +++ b/drivers/regulator/ab8500-ext.c
>> @@ -192,6 +192,10 @@ static int ab8500_ext_regulator_set_mode(struct
>> regulator_dev *rdev,
>>                 info->update_val = info->update_val_hp;
>>                 break;
>>         case REGULATOR_MODE_IDLE:
>> +               /* Always on with high power mode if info->cfg->hwreq is
>> set */
>> +               if (info->cfg && info->cfg->hwreq)
>> +                       return -EINVAL;
>> +
>>                 info->update_val = info->update_val_lp;
>>                 break;
>>
>
> I prefer to have info->update_val updated to reflect requested mode (in
> get_mode) instead of returning -EINVAL. This is how I interpret Mark's
> comment on the other patch ([PATCH] regulator: ab8500: Fix get_mode for
> shared mode regulators).Otherwise, the patch should work fine.

Well, I drop this path and send v2 for other patches now.

Thanks for the review,
Axel