[PATCH] soundwire: qcom: use signed variable for error return

[PATCH] soundwire: qcom: use signed variable for error return

[Vinod Koul]

We get warning for using a unsigned variable being compared to less than
zero. The comparison is correct as it checks for errors from previous
call to qcom_swrm_get_alert_slave_dev_num(), so we should use a signed
variable instead.

drivers/soundwire/qcom.c: qcom_swrm_irq_handler() warn: impossible
condition '(devnum < 0) => (0-255 < 0)'

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
---
 drivers/soundwire/qcom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c
index b08ecb9b418c..55ed133c6704 100644
--- a/drivers/soundwire/qcom.c
+++ b/drivers/soundwire/qcom.c
@@ -428,7 +428,7 @@ static irqreturn_t qcom_swrm_irq_handler(int irq, void *dev_id)
 	struct qcom_swrm_ctrl *swrm = dev_id;
 	u32 value, intr_sts, intr_sts_masked, slave_status;
 	u32 i;
-	u8 devnum = 0;
+	s8 devnum = 0;
 	int ret = IRQ_HANDLED;
 
 	swrm->reg_read(swrm, SWRM_INTERRUPT_STATUS, &intr_sts);
-- 
2.26.3

Re: [PATCH] soundwire: qcom: use signed variable for error return

[Pierre-Louis Bossart]

On 3/31/21 2:21 AM, Vinod Koul wrote:
> We get warning for using a unsigned variable being compared to less than
> zero. The comparison is correct as it checks for errors from previous
> call to qcom_swrm_get_alert_slave_dev_num(), so we should use a signed
> variable instead.
> 
> drivers/soundwire/qcom.c: qcom_swrm_irq_handler() warn: impossible
> condition '(devnum < 0) => (0-255 < 0)'
> 
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Vinod Koul <vkoul@kernel.org>
> ---
>   drivers/soundwire/qcom.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c
> index b08ecb9b418c..55ed133c6704 100644
> --- a/drivers/soundwire/qcom.c
> +++ b/drivers/soundwire/qcom.c
> @@ -428,7 +428,7 @@ static irqreturn_t qcom_swrm_irq_handler(int irq, void *dev_id)
>   	struct qcom_swrm_ctrl *swrm = dev_id;
>   	u32 value, intr_sts, intr_sts_masked, slave_status;
>   	u32 i;
> -	u8 devnum = 0;
> +	s8 devnum = 0;

it's not great to store negative error codes with s8. That works in this 
specific case because the function only returns -EINVAL.

We actually have zero occurrences of s8 in the drivers/soundwire/ code.

>   	int ret = IRQ_HANDLED;
>   
>   	swrm->reg_read(swrm, SWRM_INTERRUPT_STATUS, &intr_sts);
> 

Re: [PATCH] soundwire: qcom: use signed variable for error return

[Vinod Koul]

On 31-03-21, 09:41, Pierre-Louis Bossart wrote:
> 
> 
> On 3/31/21 2:21 AM, Vinod Koul wrote:
> > We get warning for using a unsigned variable being compared to less than
> > zero. The comparison is correct as it checks for errors from previous
> > call to qcom_swrm_get_alert_slave_dev_num(), so we should use a signed
> > variable instead.
> > 
> > drivers/soundwire/qcom.c: qcom_swrm_irq_handler() warn: impossible
> > condition '(devnum < 0) => (0-255 < 0)'
> > 
> > Reported-by: kernel test robot <lkp@intel.com>
> > Signed-off-by: Vinod Koul <vkoul@kernel.org>
> > ---
> >   drivers/soundwire/qcom.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c
> > index b08ecb9b418c..55ed133c6704 100644
> > --- a/drivers/soundwire/qcom.c
> > +++ b/drivers/soundwire/qcom.c
> > @@ -428,7 +428,7 @@ static irqreturn_t qcom_swrm_irq_handler(int irq, void *dev_id)
> >   	struct qcom_swrm_ctrl *swrm = dev_id;
> >   	u32 value, intr_sts, intr_sts_masked, slave_status;
> >   	u32 i;
> > -	u8 devnum = 0;
> > +	s8 devnum = 0;
> 
> it's not great to store negative error codes with s8. That works in this
> specific case because the function only returns -EINVAL.

Yeah I did check EINVAL was the case which would work but in general I
agree that makes sense, I discussed with Srini on IRC and looks like I
havent posted v2, should hit the pipes shortly

-- 
~Vinod

Re: [PATCH] soundwire: qcom: use signed variable for error return

[Bjorn Andersson]

On Wed 31 Mar 02:21 CDT 2021, Vinod Koul wrote:

> We get warning for using a unsigned variable being compared to less than
> zero. The comparison is correct as it checks for errors from previous
> call to qcom_swrm_get_alert_slave_dev_num(), so we should use a signed
> variable instead.
> 
> drivers/soundwire/qcom.c: qcom_swrm_irq_handler() warn: impossible
> condition '(devnum < 0) => (0-255 < 0)'
> 
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Vinod Koul <vkoul@kernel.org>
> ---
>  drivers/soundwire/qcom.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c
> index b08ecb9b418c..55ed133c6704 100644
> --- a/drivers/soundwire/qcom.c
> +++ b/drivers/soundwire/qcom.c
> @@ -428,7 +428,7 @@ static irqreturn_t qcom_swrm_irq_handler(int irq, void *dev_id)
>  	struct qcom_swrm_ctrl *swrm = dev_id;
>  	u32 value, intr_sts, intr_sts_masked, slave_status;
>  	u32 i;
> -	u8 devnum = 0;
> +	s8 devnum = 0;

At least in today's linux-next qcom_swrm_get_alert_slave_dev_num()
returns an int and the code only checks to see if this is negative.  So
it seems like making this a full int ensures there's no truncation etc.

And at least as written today there's no need to initialize the
variable.

Regards,
Bjorn

>  	int ret = IRQ_HANDLED;
>  
>  	swrm->reg_read(swrm, SWRM_INTERRUPT_STATUS, &intr_sts);
> -- 
> 2.26.3
>