[PATCH] media: venus: use div64_u64() instead of do_div()

[PATCH] media: venus: use div64_u64() instead of do_div()

[Himanshu Bhavani]

do_div() does a 64-by-32 division.
When the divisor is u64, do_div() truncates it to 32 bits, this means it
can test non-zero and be truncated to zero for division.

fix do_div.cocci warning:
do_div() does a 64-by-32 division, please consider using div64_u64 instead.

Signed-off-by: Himanshu Bhavani <himanshu.bhavani@siliconsignals.io>

diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
index 44b13696cf82..81853eb2993a 100644
--- a/drivers/media/platform/qcom/venus/venc.c
+++ b/drivers/media/platform/qcom/venus/venc.c
@@ -409,13 +409,13 @@ static int venc_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
 	out->capability = V4L2_CAP_TIMEPERFRAME;
 
 	us_per_frame = timeperframe->numerator * (u64)USEC_PER_SEC;
-	do_div(us_per_frame, timeperframe->denominator);
+	div64_u64(us_per_frame, timeperframe->denominator);
 
 	if (!us_per_frame)
 		return -EINVAL;
 
 	fps = (u64)USEC_PER_SEC;
-	do_div(fps, us_per_frame);
+	div64_u64(fps, us_per_frame);
 
 	inst->timeperframe = *timeperframe;
 	inst->fps = fps;
-- 
2.25.1

Re: [PATCH] media: venus: use div64_u64() instead of do_div()

[Dikshita Agarwal]

On 12/29/2023 5:45 PM, Himanshu Bhavani wrote:
> do_div() does a 64-by-32 division.
> When the divisor is u64, do_div() truncates it to 32 bits, this means it
> can test non-zero and be truncated to zero for division.
> 
> fix do_div.cocci warning:
> do_div() does a 64-by-32 division, please consider using div64_u64 instead.
> 
> Signed-off-by: Himanshu Bhavani <himanshu.bhavani@siliconsignals.io>
> 
> diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
> index 44b13696cf82..81853eb2993a 100644
> --- a/drivers/media/platform/qcom/venus/venc.c
> +++ b/drivers/media/platform/qcom/venus/venc.c
> @@ -409,13 +409,13 @@ static int venc_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
>  	out->capability = V4L2_CAP_TIMEPERFRAME;
>  
>  	us_per_frame = timeperframe->numerator * (u64)USEC_PER_SEC;
> -	do_div(us_per_frame, timeperframe->denominator);
> +	div64_u64(us_per_frame, timeperframe->denominator);
I think, this should be like below:
us_per_frame = div64_u64(us_per_frame, timeperframe->denominator);

because do_div does the in-place division while div64_u64 doesn't.
so final value of us_per_frame won't be correct with this change.

>  
>  	if (!us_per_frame)
>  		return -EINVAL;
>  
>  	fps = (u64)USEC_PER_SEC;
> -	do_div(fps, us_per_frame);
> +	div64_u64(fps, us_per_frame);
ditto..
fps = div64_u64(fps, us_per_frame);
>  
>  	inst->timeperframe = *timeperframe;
>  	inst->fps = fps;

Thanks,
Dikshita

Re: [PATCH] media: venus: use div64_u64() instead of do_div()

[Dmitry Baryshkov]

On Fri, 29 Dec 2023 at 14:16, Himanshu Bhavani
<himanshu.bhavani@siliconsignals.io> wrote:
>
> do_div() does a 64-by-32 division.
> When the divisor is u64, do_div() truncates it to 32 bits, this means it
> can test non-zero and be truncated to zero for division.
>
> fix do_div.cocci warning:
> do_div() does a 64-by-32 division, please consider using div64_u64 instead.
>
> Signed-off-by: Himanshu Bhavani <himanshu.bhavani@siliconsignals.io>
>
> diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
> index 44b13696cf82..81853eb2993a 100644
> --- a/drivers/media/platform/qcom/venus/venc.c
> +++ b/drivers/media/platform/qcom/venus/venc.c
> @@ -409,13 +409,13 @@ static int venc_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
>         out->capability = V4L2_CAP_TIMEPERFRAME;
>
>         us_per_frame = timeperframe->numerator * (u64)USEC_PER_SEC;
> -       do_div(us_per_frame, timeperframe->denominator);
> +       div64_u64(us_per_frame, timeperframe->denominator);

NAK! This is completely incorrect. do_div() is a macro and it changes
the first argument. div64_u64 is a function, which returns the result
instead of changing the first argument.

Please consider checking the code before sending a patch.

>
>         if (!us_per_frame)
>                 return -EINVAL;
>
>         fps = (u64)USEC_PER_SEC;
> -       do_div(fps, us_per_frame);
> +       div64_u64(fps, us_per_frame);
>
>         inst->timeperframe = *timeperframe;
>         inst->fps = fps;
> --
> 2.25.1
>
>


-- 
With best wishes
Dmitry