hwmon:fam15h_power Use div64_u64() for 64-bit divisor

[PATCH] hwmon:fam15h_power Use div64_u64() for 64-bit divisor

Posted by SeungJu Cheon 1 month, 1 week ago

tdelta is u64, but do_div() truncates the divisor to 32 bits.
Use div64_u64() to handle the full 64-bit divisor correctly.

Signed-off-by: SeungJu Cheon <suunj1331@gmail.com>
---
 drivers/hwmon/fam15h_power.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/hwmon/fam15h_power.c b/drivers/hwmon/fam15h_power.c
index 8ecebea53651..5e3692606516 100644
--- a/drivers/hwmon/fam15h_power.c
+++ b/drivers/hwmon/fam15h_power.c
@@ -241,7 +241,7 @@ static ssize_t power1_average_show(struct device *dev,
 		}
 		tdelta = data->cpu_sw_pwr_ptsc[cu] - prev_ptsc[cu];
 		jdelta[cu] *= data->cpu_pwr_sample_ratio * 1000;
-		do_div(jdelta[cu], tdelta);
+		jdelta[cu] = div64_u64(jdelta[cu], tdelta);
 
 		/* the unit is microWatt */
 		avg_acc += jdelta[cu];
-- 
2.52.0

Re: [PATCH] hwmon:fam15h_power Use div64_u64() for 64-bit divisor

Posted by David Laight 1 month, 1 week ago

On Tue, 30 Dec 2025 21:09:59 +0900
SeungJu Cheon <suunj1331@gmail.com> wrote:

> tdelta is u64, but do_div() truncates the divisor to 32 bits.
> Use div64_u64() to handle the full 64-bit divisor correctly.

Looking at the code I think that tdelta is a time interval and will
always fit in 32bits - so the code is probably fine.

Also I can't see anything that requires jdelta[] be an array.
Neither can I see the justification for MAX_CUS being 8.

> 
> Signed-off-by: SeungJu Cheon <suunj1331@gmail.com>
> ---
>  drivers/hwmon/fam15h_power.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/hwmon/fam15h_power.c b/drivers/hwmon/fam15h_power.c
> index 8ecebea53651..5e3692606516 100644
> --- a/drivers/hwmon/fam15h_power.c
> +++ b/drivers/hwmon/fam15h_power.c
> @@ -241,7 +241,7 @@ static ssize_t power1_average_show(struct device *dev,
>  		}
>  		tdelta = data->cpu_sw_pwr_ptsc[cu] - prev_ptsc[cu];
>  		jdelta[cu] *= data->cpu_pwr_sample_ratio * 1000;
> -		do_div(jdelta[cu], tdelta);
> +		jdelta[cu] = div64_u64(jdelta[cu], tdelta);
>  
>  		/* the unit is microWatt */
>  		avg_acc += jdelta[cu];

That could be:
		jdelta *= data->cpu_pwr_sample_ratio * 1000;
		avg_acc += div64_u64(jdelta, tdelta);
(or even 1 line...)

	David

Re: [PATCH] hwmon:fam15h_power Use div64_u64() for 64-bit divisor

Posted by Guenter Roeck 1 month, 1 week ago

On Tue, Dec 30, 2025 at 11:23:57PM +0000, David Laight wrote:
> On Tue, 30 Dec 2025 21:09:59 +0900
> SeungJu Cheon <suunj1331@gmail.com> wrote:
> 
> > tdelta is u64, but do_div() truncates the divisor to 32 bits.
> > Use div64_u64() to handle the full 64-bit divisor correctly.
> 
> Looking at the code I think that tdelta is a time interval and will
> always fit in 32bits - so the code is probably fine.
> 
Agreed.

> Also I can't see anything that requires jdelta[] be an array.
> Neither can I see the justification for MAX_CUS being 8.
> 
Also agree.

Guenter

> > 
> > Signed-off-by: SeungJu Cheon <suunj1331@gmail.com>
> > ---
> >  drivers/hwmon/fam15h_power.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/drivers/hwmon/fam15h_power.c b/drivers/hwmon/fam15h_power.c
> > index 8ecebea53651..5e3692606516 100644
> > --- a/drivers/hwmon/fam15h_power.c
> > +++ b/drivers/hwmon/fam15h_power.c
> > @@ -241,7 +241,7 @@ static ssize_t power1_average_show(struct device *dev,
> >  		}
> >  		tdelta = data->cpu_sw_pwr_ptsc[cu] - prev_ptsc[cu];
> >  		jdelta[cu] *= data->cpu_pwr_sample_ratio * 1000;
> > -		do_div(jdelta[cu], tdelta);
> > +		jdelta[cu] = div64_u64(jdelta[cu], tdelta);
> >  
> >  		/* the unit is microWatt */
> >  		avg_acc += jdelta[cu];
> 
> That could be:
> 		jdelta *= data->cpu_pwr_sample_ratio * 1000;
> 		avg_acc += div64_u64(jdelta, tdelta);
> (or even 1 line...)
> 
> 	David
> 
>