[PATCH] cpufreq/scpi: fix handling return value of topology_physical_package_id

[PATCH] cpufreq/scpi: fix handling return value of topology_physical_package_id

[Andrzej Hajda]

The function can return negative values, so its result should
be assigned to signed variable.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
---
 drivers/cpufreq/scpi-cpufreq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/cpufreq/scpi-cpufreq.c b/drivers/cpufreq/scpi-cpufreq.c
index 2c3b16f..de5e89b 100644
--- a/drivers/cpufreq/scpi-cpufreq.c
+++ b/drivers/cpufreq/scpi-cpufreq.c
@@ -31,7 +31,7 @@ static struct scpi_ops *scpi_ops;
 
 static struct scpi_dvfs_info *scpi_get_dvfs_info(struct device *cpu_dev)
 {
-	u8 domain = topology_physical_package_id(cpu_dev->id);
+	int domain = topology_physical_package_id(cpu_dev->id);
 
 	if (domain < 0)
 		return ERR_PTR(-EINVAL);
-- 
1.9.1

[PATCH] pinctrl: nsp-gpio: fix type of iterator

[Andrzej Hajda]

Iterator i declared as unsigned is always non-negative so the
loop will never end.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
---
 drivers/pinctrl/bcm/pinctrl-nsp-gpio.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
index 34648f6..ad5b04c 100644
--- a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
+++ b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
@@ -439,7 +439,8 @@ static int nsp_gpio_set_strength(struct nsp_gpio *chip, unsigned gpio,
 static int nsp_gpio_get_strength(struct nsp_gpio *chip, unsigned gpio,
 				 u16 *strength)
 {
-	unsigned int i, offset, shift;
+	unsigned int offset, shift;
+	int i;
 	u32 val;
 	unsigned long flags;
 
-- 
1.9.1

[PATCH] NFC: nci: fix handling return value of nci_hci_create_pipe

[Andrzej Hajda]

The function return NCI_HCI_INVALID_PIPE in case of error.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
---
 net/nfc/nci/hci.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/nfc/nci/hci.c b/net/nfc/nci/hci.c
index 2aedac1..a0ab26d 100644
--- a/net/nfc/nci/hci.c
+++ b/net/nfc/nci/hci.c
@@ -676,7 +676,7 @@ int nci_hci_connect_gate(struct nci_dev *ndev,
 	break;
 	default:
 		pipe = nci_hci_create_pipe(ndev, dest_host, dest_gate, &r);
-		if (pipe < 0)
+		if (pipe == NCI_HCI_INVALID_PIPE)
 			return r;
 		pipe_created = true;
 		break;
-- 
1.9.1

[PATCH] ASoC: rsnd: fix usrcnt decrementing bug

[Andrzej Hajda]

Field usrcnt is unsigned so it cannot be lesser than zero.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
---
 sound/soc/sh/rcar/ssi.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/sound/soc/sh/rcar/ssi.c b/sound/soc/sh/rcar/ssi.c
index 0b91692..8ca30fd 100644
--- a/sound/soc/sh/rcar/ssi.c
+++ b/sound/soc/sh/rcar/ssi.c
@@ -381,9 +381,9 @@ rsnd_ssi_quit_end:
 
 	rsnd_mod_power_off(mod);
 
-	ssi->usrcnt--;
-
-	if (ssi->usrcnt < 0)
+	if (ssi->usrcnt > 0)
+		ssi->usrcnt--;
+	else
 		dev_err(dev, "%s[%d] usrcnt error\n",
 			rsnd_mod_name(mod), rsnd_mod_id(mod));
 
-- 
1.9.1

Re: [PATCH] cpufreq/scpi: fix handling return value of topology_physical_package_id

[Viresh Kumar]

On 23-12-15, 11:37, Andrzej Hajda wrote:
> The function can return negative values, so its result should
> be assigned to signed variable.
> 
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
> 
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
> 
> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
> ---
>  drivers/cpufreq/scpi-cpufreq.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/cpufreq/scpi-cpufreq.c b/drivers/cpufreq/scpi-cpufreq.c
> index 2c3b16f..de5e89b 100644
> --- a/drivers/cpufreq/scpi-cpufreq.c
> +++ b/drivers/cpufreq/scpi-cpufreq.c
> @@ -31,7 +31,7 @@ static struct scpi_ops *scpi_ops;
>  
>  static struct scpi_dvfs_info *scpi_get_dvfs_info(struct device *cpu_dev)
>  {
> -	u8 domain = topology_physical_package_id(cpu_dev->id);
> +	int domain = topology_physical_package_id(cpu_dev->id);
>  
>  	if (domain < 0)
>  		return ERR_PTR(-EINVAL);

Acked-by: Viresh Kumar <viresh.kumar@linaro.org>

-- 
viresh

Re: [PATCH] cpufreq/scpi: fix handling return value of topology_physical_package_id

[Sudeep Holla]

On 23/12/15 10:37, Andrzej Hajda wrote:
> The function can return negative values, so its result should
> be assigned to signed variable.
>
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
>
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
>

There has a patch posted by Dan Carpenter [1] which I reposted[2],
but it again slipped through the cracks. I will poke Rafael again on that.

[1] 
http://lists.infradead.org/pipermail/linux-arm-kernel/2015-October/380292.html
[2] https://lkml.org/lkml/2015/12/15/219

-- 
Regards,
Sudeep

Re: [PATCH] pinctrl: nsp-gpio: fix type of iterator

[Ray Jui]

+ Reddy

On 12/23/2015 2:37 AM, Andrzej Hajda wrote:
> Iterator i declared as unsigned is always non-negative so the
> loop will never end.
>
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
>
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
>
> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
> ---
>   drivers/pinctrl/bcm/pinctrl-nsp-gpio.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
> index 34648f6..ad5b04c 100644
> --- a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
> +++ b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
> @@ -439,7 +439,8 @@ static int nsp_gpio_set_strength(struct nsp_gpio *chip, unsigned gpio,
>   static int nsp_gpio_get_strength(struct nsp_gpio *chip, unsigned gpio,
>   				 u16 *strength)
>   {
> -	unsigned int i, offset, shift;
> +	unsigned int offset, shift;
> +	int i;
>   	u32 val;
>   	unsigned long flags;
>
>

The fix is a valid fix. And at the same time it exposes other potential 
issues in the driver. I just found the loop used in _set_strength and 
_get_strength is inconsistent:

In _set_strength:

427         for (i = GPIO_DRV_STRENGTH_BITS; i > 0; i--) {

For i to start at GPIO_DRV_STRENGTH_BITS, it seems to be wrong.

428                 val = readl(chip->io_ctrl + offset);
429                 val &= ~BIT(shift);
430                 val |= ((strength >> (i-1)) & 0x1) << shift;
431                 writel(val, chip->io_ctrl + offset);
432                 offset += 4;
433         }

In _get_strength:

451         for (i = (GPIO_DRV_STRENGTH_BITS - 1); i >= 0; i--) {
452                 val = readl(chip->io_ctrl + offset) & BIT(shift);
453                 val >>= shift;
454                 *strength += (val << i);
455                 offset += 4;
456         }

Reddy, could you please review and comment?

Thanks,

Ray

Re: [PATCH] ASoC: rsnd: fix usrcnt decrementing bug

[Kuninori Morimoto]

Hi Andrzej

> Field usrcnt is unsigned so it cannot be lesser than zero.
> 
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
> 
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
> 
> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
> ---

Thank you for your patch. good catch !
I noticed current error case is not good for ssi.c
Can you agree below ?

---------
diff --git a/sound/soc/sh/rcar/ssi.c b/sound/soc/sh/rcar/ssi.c
index 7db05fd..e519e30 100644
--- a/sound/soc/sh/rcar/ssi.c
+++ b/sound/soc/sh/rcar/ssi.c
@@ -403,6 +403,12 @@ static int rsnd_ssi_quit(struct rsnd_mod *mod,
 	struct rsnd_ssi *ssi = rsnd_mod_to_ssi(mod);
 	struct device *dev = rsnd_priv_to_dev(priv);
 
+	if (!ssi->usrcnt) {
+		dev_err(dev, "%s[%d] usrcnt error\n",
+			rsnd_mod_name(mod), rsnd_mod_id(mod));
+		return -EIO;
+	}
+
 	if (rsnd_ssi_is_parent(mod, io))
 		goto rsnd_ssi_quit_end;
 
@@ -422,10 +428,6 @@ rsnd_ssi_quit_end:
 
 	ssi->usrcnt--;
 
-	if (ssi->usrcnt < 0)
-		dev_err(dev, "%s[%d] usrcnt error\n",
-			rsnd_mod_name(mod), rsnd_mod_id(mod));
-
 	return 0;
 }
 
---------


Best regards
---
Kuninori Morimoto

Re: [PATCH] pinctrl: nsp-gpio: fix type of iterator

[Yendapally Reddy Dhananjaya Reddy]

On 12/24/2015 4:05 AM, Ray Jui wrote:
> + Reddy
>
> On 12/23/2015 2:37 AM, Andrzej Hajda wrote:
>> Iterator i declared as unsigned is always non-negative so the
>> loop will never end.
>>
>> The problem has been detected using proposed semantic patch
>> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
>>
>> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
>>
>> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
>> ---
>>   drivers/pinctrl/bcm/pinctrl-nsp-gpio.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
>> index 34648f6..ad5b04c 100644
>> --- a/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
>> +++ b/drivers/pinctrl/bcm/pinctrl-nsp-gpio.c
>> @@ -439,7 +439,8 @@ static int nsp_gpio_set_strength(struct nsp_gpio *chip, unsigned gpio,
>>   static int nsp_gpio_get_strength(struct nsp_gpio *chip, unsigned gpio,
>>   				 u16 *strength)
>>   {
>> -	unsigned int i, offset, shift;
>> +	unsigned int offset, shift;
>> +	int i;
>>   	u32 val;
>>   	unsigned long flags;
>>
>>
> The fix is a valid fix. And at the same time it exposes other potential 
> issues in the driver. I just found the loop used in _set_strength and 
> _get_strength is inconsistent:
>
> In _set_strength:
>
> 427         for (i = GPIO_DRV_STRENGTH_BITS; i > 0; i--) {
>
> For i to start at GPIO_DRV_STRENGTH_BITS, it seems to be wrong.
>
> 428                 val = readl(chip->io_ctrl + offset);
> 429                 val &= ~BIT(shift);
> 430                 val |= ((strength >> (i-1)) & 0x1) << shift;
> 431                 writel(val, chip->io_ctrl + offset);
> 432                 offset += 4;
> 433         }
>
> In _get_strength:
>
> 451         for (i = (GPIO_DRV_STRENGTH_BITS - 1); i >= 0; i--) {
> 452                 val = readl(chip->io_ctrl + offset) & BIT(shift);
> 453                 val >>= shift;
> 454                 *strength += (val << i);
> 455                 offset += 4;
> 456         }
>
> Reddy, could you please review and comment?

Hi Ray,

             The logic is correct. The drive strength has three bits distributed in three registers. In one case the "-1" is in for loop initialization and the other case it is in the for loop body. The fix looks good.

Thanks
Dhananjay

> Thanks,
>
> Ray
>
>

Re: [PATCH] ASoC: rsnd: fix usrcnt decrementing bug

[Andrzej Hajda]

On 12/24/2015 01:04 AM, Kuninori Morimoto wrote:
> Hi Andrzej
>
>> Field usrcnt is unsigned so it cannot be lesser than zero.
>>
>> The problem has been detected using proposed semantic patch
>> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
>>
>> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
>>
>> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
>> ---
> Thank you for your patch. good catch !
> I noticed current error case is not good for ssi.c
> Can you agree below ?
Yes, of course.

Regards
Andrzej
>
> ---------
> diff --git a/sound/soc/sh/rcar/ssi.c b/sound/soc/sh/rcar/ssi.c
> index 7db05fd..e519e30 100644
> --- a/sound/soc/sh/rcar/ssi.c
> +++ b/sound/soc/sh/rcar/ssi.c
> @@ -403,6 +403,12 @@ static int rsnd_ssi_quit(struct rsnd_mod *mod,
>  	struct rsnd_ssi *ssi = rsnd_mod_to_ssi(mod);
>  	struct device *dev = rsnd_priv_to_dev(priv);
>  
> +	if (!ssi->usrcnt) {
> +		dev_err(dev, "%s[%d] usrcnt error\n",
> +			rsnd_mod_name(mod), rsnd_mod_id(mod));
> +		return -EIO;
> +	}
> +
>  	if (rsnd_ssi_is_parent(mod, io))
>  		goto rsnd_ssi_quit_end;
>  
> @@ -422,10 +428,6 @@ rsnd_ssi_quit_end:
>  
>  	ssi->usrcnt--;
>  
> -	if (ssi->usrcnt < 0)
> -		dev_err(dev, "%s[%d] usrcnt error\n",
> -			rsnd_mod_name(mod), rsnd_mod_id(mod));
> -
>  	return 0;
>  }
>  
> ---------
>
>
> Best regards
> ---
> Kuninori Morimoto
>
>

Re: [PATCH] ASoC: rsnd: fix usrcnt decrementing bug

[Kuninori Morimoto]

Hi Andrzej

> >> Field usrcnt is unsigned so it cannot be lesser than zero.
> >>
> >> The problem has been detected using proposed semantic patch
> >> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
> >>
> >> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
> >>
> >> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
> >> ---
> > Thank you for your patch. good catch !
> > I noticed current error case is not good for ssi.c
> > Can you agree below ?
> Yes, of course.

Thanks.
Who send this patch ? you or me ?
I think you can do it ?

[PATCH v2] ASoC: rsnd: fix usrcnt decrementing bug

[Andrzej Hajda]

Field usrcnt is unsigned so it cannot be lesser than zero.
The patch fixes the check, moves it to the beginning of the function
and changes return value to -EIO in case of usercnt error.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
---
v2: changed according to Kuninori Morimoto advice
---
 sound/soc/sh/rcar/ssi.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)

diff --git a/sound/soc/sh/rcar/ssi.c b/sound/soc/sh/rcar/ssi.c
index 0b91692..f23c921 100644
--- a/sound/soc/sh/rcar/ssi.c
+++ b/sound/soc/sh/rcar/ssi.c
@@ -364,29 +364,30 @@ static int rsnd_ssi_quit(struct rsnd_mod *mod,
 	struct rsnd_ssi *ssi = rsnd_mod_to_ssi(mod);
 	struct device *dev = rsnd_priv_to_dev(priv);
 
-	if (rsnd_ssi_is_parent(mod, io))
-		goto rsnd_ssi_quit_end;
+	if (!ssi->usrcnt) {
+		dev_err(dev, "%s[%d] usrcnt error\n",
+			rsnd_mod_name(mod), rsnd_mod_id(mod));
+		return -EIO;
+	}
 
-	if (ssi->err > 0)
-		dev_warn(dev, "%s[%d] under/over flow err = %d\n",
-			 rsnd_mod_name(mod), rsnd_mod_id(mod), ssi->err);
+	if (!rsnd_ssi_is_parent(mod, io)) {
+		if (ssi->err > 0)
+			dev_warn(dev, "%s[%d] under/over flow err = %d\n",
+				 rsnd_mod_name(mod), rsnd_mod_id(mod),
+				 ssi->err);
 
-	ssi->cr_own	= 0;
-	ssi->err	= 0;
+		ssi->cr_own	= 0;
+		ssi->err	= 0;
 
-	rsnd_ssi_irq_disable(mod);
+		rsnd_ssi_irq_disable(mod);
+	}
 
-rsnd_ssi_quit_end:
 	rsnd_ssi_master_clk_stop(ssi, io);
 
 	rsnd_mod_power_off(mod);
 
 	ssi->usrcnt--;
 
-	if (ssi->usrcnt < 0)
-		dev_err(dev, "%s[%d] usrcnt error\n",
-			rsnd_mod_name(mod), rsnd_mod_id(mod));
-
 	return 0;
 }
 
-- 
1.9.1

Re: [PATCH v2] ASoC: rsnd: fix usrcnt decrementing bug

[Kuninori Morimoto]

Hi
> 
> Field usrcnt is unsigned so it cannot be lesser than zero.
> The patch fixes the check, moves it to the beginning of the function
> and changes return value to -EIO in case of usercnt error.
> 
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
> 
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
> 
> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
> ---

Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>

Re: [PATCH] pinctrl: nsp-gpio: fix type of iterator

[Linus Walleij]

On Wed, Dec 23, 2015 at 11:37 AM, Andrzej Hajda <a.hajda@samsung.com> wrote:

> Iterator i declared as unsigned is always non-negative so the
> loop will never end.
>
> The problem has been detected using proposed semantic patch
> scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].
>
> [1]: http://permalink.gmane.org/gmane.linux.kernel/2038576
>
> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>

I already applied Dan Carpenters fix, somehow it was earlier
in my mail flow. Thanks anyways!

Yours,
Linus Walleij

Re: [PATCH v2] ASoC: rsnd: fix usrcnt decrementing bug

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 379 bytes --]

On Thu, Dec 24, 2015 at 08:02:39AM +0100, Andrzej Hajda wrote:
> Field usrcnt is unsigned so it cannot be lesser than zero.
> The patch fixes the check, moves it to the beginning of the function
> and changes return value to -EIO in case of usercnt error.

Please don't send new patches in reply to old ones, it makes it hard to
figure out what the current version of things is.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]