As the comment explains, the if check ensures that the divisor oa_period
is a u32. Explicitly cast oa_period to u32 to remove the following
Coccinelle/coccicheck warning reported by do_div.cocci:

  WARNING: do_div() does a 64-by-32 division, please consider using div64_u64 instead

Use the preferred div_u64() function instead of the do_div() macro and
remove the now unnecessary local variable tmp.

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
---
Changes in v2:
- Use div_u64() instead of do_div() after feedback from Ville Syrjälä
- Link to v1: https://lore.kernel.org/linux-kernel/20240710074650.419902-2-thorsten.blum@toblux.com/
---
 drivers/gpu/drm/i915/i915_perf.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c
index 0b1cd4c7a525..f65fbe13ab59 100644
--- a/drivers/gpu/drm/i915/i915_perf.c
+++ b/drivers/gpu/drm/i915/i915_perf.c
@@ -4096,15 +4096,13 @@ static int read_properties_unlocked(struct i915_perf *perf,
 			oa_period = oa_exponent_to_ns(perf, value);
 
 			/* This check is primarily to ensure that oa_period <=
-			 * UINT32_MAX (before passing to do_div which only
+			 * UINT32_MAX (before passing it to div_u64 which only
 			 * accepts a u32 denominator), but we can also skip
 			 * checking anything < 1Hz which implicitly can't be
 			 * limited via an integer oa_max_sample_rate.
 			 */
 			if (oa_period <= NSEC_PER_SEC) {
-				u64 tmp = NSEC_PER_SEC;
-				do_div(tmp, oa_period);
-				oa_freq_hz = tmp;
+				oa_freq_hz = div_u64(NSEC_PER_SEC, (u32)oa_period);
 			} else
 				oa_freq_hz = 0;
 
-- 
2.45.2

-----Original Message-----
From: Intel-gfx <intel-gfx-bounces@lists.freedesktop.org> On Behalf Of Thorsten Blum
Sent: Friday, August 2, 2024 9:03 AM
To: jani.nikula@linux.intel.com; joonas.lahtinen@linux.intel.com; Vivi, Rodrigo <rodrigo.vivi@intel.com>; tursulin@ursulin.net; airlied@gmail.com; daniel@ffwll.ch
Cc: intel-gfx@lists.freedesktop.org; dri-devel@lists.freedesktop.org; linux-kernel@vger.kernel.org; Thorsten Blum <thorsten.blum@toblux.com>
Subject: [PATCH v2] drm/i915: Explicitly cast divisor and use div_u64()
> 
> As the comment explains, the if check ensures that the divisor oa_period
> is a u32. Explicitly cast oa_period to u32 to remove the following
> Coccinelle/coccicheck warning reported by do_div.cocci:
> 
>   WARNING: do_div() does a 64-by-32 division, please consider using div64_u64 instead
> 
> Use the preferred div_u64() function instead of the do_div() macro and
> remove the now unnecessary local variable tmp.
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
> ---
> Changes in v2:
> - Use div_u64() instead of do_div() after feedback from Ville Syrjälä
> - Link to v1: https://lore.kernel.org/linux-kernel/20240710074650.419902-2-thorsten.blum@toblux.com/
> ---
>  drivers/gpu/drm/i915/i915_perf.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c
> index 0b1cd4c7a525..f65fbe13ab59 100644
> --- a/drivers/gpu/drm/i915/i915_perf.c
> +++ b/drivers/gpu/drm/i915/i915_perf.c
> @@ -4096,15 +4096,13 @@ static int read_properties_unlocked(struct i915_perf *perf,
>  			oa_period = oa_exponent_to_ns(perf, value);
>  
>  			/* This check is primarily to ensure that oa_period <=
> -			 * UINT32_MAX (before passing to do_div which only
> +			 * UINT32_MAX (before passing it to div_u64 which only
>  			 * accepts a u32 denominator), but we can also skip
>  			 * checking anything < 1Hz which implicitly can't be
>  			 * limited via an integer oa_max_sample_rate.
>  			 */
>  			if (oa_period <= NSEC_PER_SEC) {
> -				u64 tmp = NSEC_PER_SEC;
> -				do_div(tmp, oa_period);
> -				oa_freq_hz = tmp;
> +				oa_freq_hz = div_u64(NSEC_PER_SEC, (u32)oa_period);
>  			} else
>  				oa_freq_hz = 0;

Non-blocking suggestion: this looks like it can be inlined.  And if the
inline route is taken, it might be best to invert the conditional check
like such:

oa_freq_hz = oa_period > NSEC_PER_SEC ? 0 :
                                     div_u64(NSEC_PER_SEC, (u32)oa_period);

I think this is just a matter of preference, though.  The explicit if-else
block is definitely clearer.
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
-Jonathan Cavitt

>  
> -- 
> 2.45.2
> 
> 

Hi Thorsten,

> >  			/* This check is primarily to ensure that oa_period <=
> > -			 * UINT32_MAX (before passing to do_div which only
> > +			 * UINT32_MAX (before passing it to div_u64 which only
> >  			 * accepts a u32 denominator), but we can also skip
> >  			 * checking anything < 1Hz which implicitly can't be
> >  			 * limited via an integer oa_max_sample_rate.
> >  			 */
> >  			if (oa_period <= NSEC_PER_SEC) {
> > -				u64 tmp = NSEC_PER_SEC;
> > -				do_div(tmp, oa_period);
> > -				oa_freq_hz = tmp;
> > +				oa_freq_hz = div_u64(NSEC_PER_SEC, (u32)oa_period);
> >  			} else
> >  				oa_freq_hz = 0;
> 
> Non-blocking suggestion: this looks like it can be inlined.  And if the
> inline route is taken, it might be best to invert the conditional check
> like such:
> 
> oa_freq_hz = oa_period > NSEC_PER_SEC ? 0 :
>                                      div_u64(NSEC_PER_SEC, (u32)oa_period);
> 
> I think this is just a matter of preference, though.  The explicit if-else
> block is definitely clearer.

It's also stylistically wrong given that now the if/else don't
need the brackets anymore, triggering a checkpatch error.

Thorsten do you mind resending it either following Jonathan's
suggestion (my favourite, as well) or fix the bracket issue
following the kernel style.

Andi