[PATCH] media: i2c: adp1653: fix error handling from a call to adp1653_get_fault

[PATCH] media: i2c: adp1653: fix error handling from a call to adp1653_get_fault

[Colin King]

From: Colin Ian King <colin.king@canonical.com>

The error check on rval from the call to adp1653_get_fault currently
returns if rval is non-zero. This appears to be incorrect as the
following if statement checks for various bit settings in rval so
clearly rval is expected to be non-zero at that point. Coverity
flagged the if statement up as deadcode.  Fix this so the error
return path only occurs when rval is negative.

Addresses-Coverity: ("Logically dead code")
Fixes: 287980e49ffc ("remove lots of IS_ERR_VALUE abuses")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
---
 drivers/media/i2c/adp1653.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/media/i2c/adp1653.c b/drivers/media/i2c/adp1653.c
index 522a0b10e415..1a4878385394 100644
--- a/drivers/media/i2c/adp1653.c
+++ b/drivers/media/i2c/adp1653.c
@@ -170,7 +170,7 @@ static int adp1653_set_ctrl(struct v4l2_ctrl *ctrl)
 	int rval;
 
 	rval = adp1653_get_fault(flash);
-	if (rval)
+	if (rval < 0)
 		return rval;
 	if ((rval & (ADP1653_REG_FAULT_FLT_SCP |
 		     ADP1653_REG_FAULT_FLT_OT |
-- 
2.30.0

Re: [PATCH] media: i2c: adp1653: fix error handling from a call to adp1653_get_fault

[Dan Carpenter]

On Fri, Feb 26, 2021 at 11:22:29PM +0000, Colin King wrote:
> From: Colin Ian King <colin.king@canonical.com>
> 
> The error check on rval from the call to adp1653_get_fault currently
> returns if rval is non-zero. This appears to be incorrect as the
> following if statement checks for various bit settings in rval so
> clearly rval is expected to be non-zero at that point. Coverity
> flagged the if statement up as deadcode.  Fix this so the error
> return path only occurs when rval is negative.
> 
> Addresses-Coverity: ("Logically dead code")
> Fixes: 287980e49ffc ("remove lots of IS_ERR_VALUE abuses")
> Signed-off-by: Colin Ian King <colin.king@canonical.com>
> ---
>  drivers/media/i2c/adp1653.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/media/i2c/adp1653.c b/drivers/media/i2c/adp1653.c
> index 522a0b10e415..1a4878385394 100644
> --- a/drivers/media/i2c/adp1653.c
> +++ b/drivers/media/i2c/adp1653.c
> @@ -170,7 +170,7 @@ static int adp1653_set_ctrl(struct v4l2_ctrl *ctrl)
>  	int rval;
>  
>  	rval = adp1653_get_fault(flash);
> -	if (rval)
> +	if (rval < 0)
>  		return rval;

This is good, but all the other callers need to fixed as well:


   140  static int adp1653_get_ctrl(struct v4l2_ctrl *ctrl)
   141  {
   142          struct adp1653_flash *flash =
   143                  container_of(ctrl->handler, struct adp1653_flash, ctrls);
   144          int rval;
   145  
   146          rval = adp1653_get_fault(flash);
   147          if (rval)
   148                  return rval;
   149  
   150          ctrl->cur.val = 0;
   151  
   152          if (flash->fault & ADP1653_REG_FAULT_FLT_SCP)
                    ^^^^^^^^^^^^
flash->fault is the equivalent of "rval" for non-negative returns so
this condition can never be true.  We should never be returning these
weird firmware ADP1653_REG_FAULT_FLT_SCP fault codes to the v4l2 layers.

   153                  ctrl->cur.val |= V4L2_FLASH_FAULT_SHORT_CIRCUIT;
   154          if (flash->fault & ADP1653_REG_FAULT_FLT_OT)
   155                  ctrl->cur.val |= V4L2_FLASH_FAULT_OVER_TEMPERATURE;
   156          if (flash->fault & ADP1653_REG_FAULT_FLT_TMR)
   157                  ctrl->cur.val |= V4L2_FLASH_FAULT_TIMEOUT;
   158          if (flash->fault & ADP1653_REG_FAULT_FLT_OV)
   159                  ctrl->cur.val |= V4L2_FLASH_FAULT_OVER_VOLTAGE;
   160  
   161          flash->fault = 0;
   162  
   163          return 0;
   164  }

regards,
dan carpenter

Re: [PATCH] media: i2c: adp1653: fix error handling from a call to adp1653_get_fault

[Sakari Ailus]

Hi Dan, Colin,

On Sat, Feb 27, 2021 at 01:17:20PM +0300, Dan Carpenter wrote:
> On Fri, Feb 26, 2021 at 11:22:29PM +0000, Colin King wrote:
> > From: Colin Ian King <colin.king@canonical.com>
> > 
> > The error check on rval from the call to adp1653_get_fault currently
> > returns if rval is non-zero. This appears to be incorrect as the
> > following if statement checks for various bit settings in rval so
> > clearly rval is expected to be non-zero at that point. Coverity
> > flagged the if statement up as deadcode.  Fix this so the error
> > return path only occurs when rval is negative.
> > 
> > Addresses-Coverity: ("Logically dead code")
> > Fixes: 287980e49ffc ("remove lots of IS_ERR_VALUE abuses")
> > Signed-off-by: Colin Ian King <colin.king@canonical.com>
> > ---
> >  drivers/media/i2c/adp1653.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/drivers/media/i2c/adp1653.c b/drivers/media/i2c/adp1653.c
> > index 522a0b10e415..1a4878385394 100644
> > --- a/drivers/media/i2c/adp1653.c
> > +++ b/drivers/media/i2c/adp1653.c
> > @@ -170,7 +170,7 @@ static int adp1653_set_ctrl(struct v4l2_ctrl *ctrl)
> >  	int rval;
> >  
> >  	rval = adp1653_get_fault(flash);
> > -	if (rval)
> > +	if (rval < 0)
> >  		return rval;
> 
> This is good, but all the other callers need to fixed as well:
> 
> 
>    140  static int adp1653_get_ctrl(struct v4l2_ctrl *ctrl)
>    141  {
>    142          struct adp1653_flash *flash =
>    143                  container_of(ctrl->handler, struct adp1653_flash, ctrls);
>    144          int rval;
>    145  
>    146          rval = adp1653_get_fault(flash);
>    147          if (rval)
>    148                  return rval;
>    149  
>    150          ctrl->cur.val = 0;
>    151  
>    152          if (flash->fault & ADP1653_REG_FAULT_FLT_SCP)
>                     ^^^^^^^^^^^^
> flash->fault is the equivalent of "rval" for non-negative returns so
> this condition can never be true.  We should never be returning these
> weird firmware ADP1653_REG_FAULT_FLT_SCP fault codes to the v4l2 layers.

I think this could be fixed and cleaned up by always retuning zero on
success, and checking for flash->faults while holding the mutex in
adp1653_init_device. I could fix that, too, just let me know...

-- 
Regards,

Sakari Ailus