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

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

[Qing Wang]

From: Wang Qing <wangqing@vivo.com>

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: Wang Qing <wangqing@vivo.com>
---
 drivers/media/platform/qcom/venus/vdec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
index 91da3f5..941a904
--- a/drivers/media/platform/qcom/venus/vdec.c
+++ b/drivers/media/platform/qcom/venus/vdec.c
@@ -435,7 +435,7 @@ static int vdec_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
 		return -EINVAL;
 
 	fps = (u64)USEC_PER_SEC;
-	do_div(fps, us_per_frame);
+	div64_u64(fps, us_per_frame);
 
 	inst->fps = fps;
 	inst->timeperframe = *timeperframe;
-- 
2.7.4

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

[Dmitry Baryshkov]

On 09/02/2022 11:38, Qing Wang wrote:
> From: Wang Qing <wangqing@vivo.com>
> 
> 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.


NAK

The div64_u64 is not equivalent to do_div. It returns the quotient
rather than modifying the first arg. Moreover it is unoptimal on 32-bit
arches.

> 
> Signed-off-by: Wang Qing <wangqing@vivo.com>
> ---
>   drivers/media/platform/qcom/venus/vdec.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
> index 91da3f5..941a904
> --- a/drivers/media/platform/qcom/venus/vdec.c
> +++ b/drivers/media/platform/qcom/venus/vdec.c
> @@ -435,7 +435,7 @@ static int vdec_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
>   		return -EINVAL;
>   
>   	fps = (u64)USEC_PER_SEC;
> -	do_div(fps, us_per_frame);
> +	div64_u64(fps, us_per_frame);
>   
>   	inst->fps = fps;
>   	inst->timeperframe = *timeperframe;


-- 
With best wishes
Dmitry

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

[David Laight]

From: Dmitry Baryshkov
> Sent: 10 February 2022 10:17
> 
> On 09/02/2022 11:38, Qing Wang wrote:
> > From: Wang Qing <wangqing@vivo.com>
> >
> > 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.
> 
> 
> NAK
> 
> The div64_u64 is not equivalent to do_div. It returns the quotient
> rather than modifying the first arg. Moreover it is unoptimal on 32-bit
> arches.
> 
> >
> > Signed-off-by: Wang Qing <wangqing@vivo.com>
> > ---
> >   drivers/media/platform/qcom/venus/vdec.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
> > index 91da3f5..941a904
> > --- a/drivers/media/platform/qcom/venus/vdec.c
> > +++ b/drivers/media/platform/qcom/venus/vdec.c
> > @@ -435,7 +435,7 @@ static int vdec_s_parm(struct file *file, void *fh, struct v4l2_streamparm *a)
> >   		return -EINVAL;
> >
> >   	fps = (u64)USEC_PER_SEC;
> > -	do_div(fps, us_per_frame);
> > +	div64_u64(fps, us_per_frame);

That one is particularly insane to begin with.
Why was it even do_div()?
I guess because:
	fps = USEC_PER_SEC / us_per_frame;
caused the compiler to call the 64bit divide function because the 'us_per_frame'
is 64bit - even though the code only makes sense if it is small.

The correct fix here is probably to change the type of us_per_frame to 32bit
or just do:
	fps = USEC_PER_FRAM / (unsigned int)us_per_frame.

Oh, and fps definitely doesn't need to be 64bit either.

IMHO all (and I mean ALL) of these patches to change do_div() to div64_u64()
need dropping.
Maybe one or two are real problems, but the surrounding code needs
correctly examining.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)