The cpufreq documentation suggests avoiding direct pointer subtraction
when working with entries in driver_freq_table, as it is relatively
costly. Instead, the recommended approach is to use the provided
iteration macros:

- cpufreq_for_each_valid_entry_idx()

Replace pointer subtraction in freq_table.c with the macros
cpufreq_for_each_entry_idx() and cpufreq_for_each_valid_entry_idx(), as
the index does not need initialization, avoiding unnecessary
computation. This improves code clarity and follows the established
cpufreq coding style.

No functional change intended.

Signed-off-by: Zihuan Zhang <zhangzihuan@kylinos.cn>

V2:
 - Remove unnecessary initialization for current and remaining follow Rafael's suggestion
---
 drivers/cpufreq/freq_table.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c
index d5111ee56e38..408fd8fee2e3 100644
--- a/drivers/cpufreq/freq_table.c
+++ b/drivers/cpufreq/freq_table.c
@@ -33,16 +33,17 @@ int cpufreq_frequency_table_cpuinfo(struct cpufreq_policy *policy)
 	struct cpufreq_frequency_table *pos, *table = policy->freq_table;
 	unsigned int min_freq = ~0;
 	unsigned int max_freq = 0;
+	unsigned int i;
 	unsigned int freq;
 
-	cpufreq_for_each_valid_entry(pos, table) {
+	cpufreq_for_each_valid_entry_idx(pos, table, i) {
 		freq = pos->frequency;
 
 		if ((!cpufreq_boost_enabled() || !policy->boost_enabled)
 		    && (pos->flags & CPUFREQ_BOOST_FREQ))
 			continue;
 
-		pr_debug("table entry %u: %u kHz\n", (int)(pos - table), freq);
+		pr_debug("table entry %u: %u kHz\n", i, freq);
 		if (freq < min_freq)
 			min_freq = freq;
 		if (freq > max_freq)
@@ -126,7 +127,7 @@ int cpufreq_table_index_unsorted(struct cpufreq_policy *policy,
 	};
 	struct cpufreq_frequency_table *pos;
 	struct cpufreq_frequency_table *table = policy->freq_table;
-	unsigned int freq, diff, i = 0;
+	unsigned int freq, diff, i;
 	int index;
 
 	pr_debug("request for target %u kHz (relation: %u) for cpu %u\n",
-- 
2.25.1

On Sep 23, 2025 at 15:55:53 +0800, Zihuan Zhang wrote:
> The cpufreq documentation suggests avoiding direct pointer subtraction
> when working with entries in driver_freq_table, as it is relatively
> costly. Instead, the recommended approach is to use the provided
> iteration macros:

Thanks for the patch,
Just say "macro" not "macros".

> 
> - cpufreq_for_each_valid_entry_idx()
> 
> Replace pointer subtraction in freq_table.c with the macros
> cpufreq_for_each_entry_idx() and cpufreq_for_each_valid_entry_idx(), as

Where is "cpufreq_for_each_entry_idx" in this entire patch?
Please drop this reference, why confuse people?

> the index does not need initialization, avoiding unnecessary
> computation. This improves code clarity and follows the established
> cpufreq coding style.

You don't need to add all this to the commit message about the
unnecessary computation, code clarity, etc..
Please keep it to the point.

> 
> No functional change intended.
> 
> Signed-off-by: Zihuan Zhang <zhangzihuan@kylinos.cn>
> 
> V2:
>  - Remove unnecessary initialization for current and remaining follow Rafael's suggestion

You didn't fix Rafael's first comment [1] about the $subject, and also please
add links to previous revisions for ease of review.

[1] > In the subject, this is just one macro, not multiple macros.


> ---
>  drivers/cpufreq/freq_table.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c
> index d5111ee56e38..408fd8fee2e3 100644
> --- a/drivers/cpufreq/freq_table.c
> +++ b/drivers/cpufreq/freq_table.c
> @@ -33,16 +33,17 @@ int cpufreq_frequency_table_cpuinfo(struct cpufreq_policy *policy)
>  	struct cpufreq_frequency_table *pos, *table = policy->freq_table;
>  	unsigned int min_freq = ~0;
>  	unsigned int max_freq = 0;
> +	unsigned int i;
>  	unsigned int freq;
>  
> -	cpufreq_for_each_valid_entry(pos, table) {
> +	cpufreq_for_each_valid_entry_idx(pos, table, i) {
>  		freq = pos->frequency;
>  
>  		if ((!cpufreq_boost_enabled() || !policy->boost_enabled)
>  		    && (pos->flags & CPUFREQ_BOOST_FREQ))
>  			continue;
>  
> -		pr_debug("table entry %u: %u kHz\n", (int)(pos - table), freq);
> +		pr_debug("table entry %u: %u kHz\n", i, freq);
>  		if (freq < min_freq)
>  			min_freq = freq;
>  		if (freq > max_freq)
> @@ -126,7 +127,7 @@ int cpufreq_table_index_unsorted(struct cpufreq_policy *policy,
>  	};
>  	struct cpufreq_frequency_table *pos;
>  	struct cpufreq_frequency_table *table = policy->freq_table;
> -	unsigned int freq, diff, i = 0;
> +	unsigned int freq, diff, i;
>  	int index;

Usually, it isn't advised to touch code that's not strictly relevant to
your patch. However since this seems like a minor fixup it's fine by
me... Upto Rafael whether he prefers to have/not have this unrelated change.

But atleast mention in your commit message that you're also removing the
initialization from cpufreq_table_index_unsorted as part of some minor cleanup
(which seems kinda unnecessary to me TBH in the first place)

-- 
Best regards,
Dhruva Gole
Texas Instruments Incorporated

在 2025/9/23 17:09, Dhruva Gole 写道:
>
> Usually, it isn't advised to touch code that's not strictly relevant to
> your patch. However since this seems like a minor fixup it's fine by
> me... Upto Rafael whether he prefers to have/not have this unrelated change.
>
> But atleast mention in your commit message that you're also removing the
> initialization from cpufreq_table_index_unsorted as part of some minor cleanup
> (which seems kinda unnecessary to me TBH in the first place)
>

Thank you all ！

On Tue, Sep 23, 2025 at 11:09 AM Dhruva Gole <d-gole@ti.com> wrote:
>
> On Sep 23, 2025 at 15:55:53 +0800, Zihuan Zhang wrote:
> > The cpufreq documentation suggests avoiding direct pointer subtraction
> > when working with entries in driver_freq_table, as it is relatively
> > costly. Instead, the recommended approach is to use the provided
> > iteration macros:
>
> Thanks for the patch,
> Just say "macro" not "macros".
>
> >
> > - cpufreq_for_each_valid_entry_idx()
> >
> > Replace pointer subtraction in freq_table.c with the macros
> > cpufreq_for_each_entry_idx() and cpufreq_for_each_valid_entry_idx(), as
>
> Where is "cpufreq_for_each_entry_idx" in this entire patch?
> Please drop this reference, why confuse people?
>
> > the index does not need initialization, avoiding unnecessary
> > computation. This improves code clarity and follows the established
> > cpufreq coding style.
>
> You don't need to add all this to the commit message about the
> unnecessary computation, code clarity, etc..
> Please keep it to the point.
>
> >
> > No functional change intended.
> >
> > Signed-off-by: Zihuan Zhang <zhangzihuan@kylinos.cn>
> >
> > V2:
> >  - Remove unnecessary initialization for current and remaining follow Rafael's suggestion
>
> You didn't fix Rafael's first comment [1] about the $subject, and also please
> add links to previous revisions for ease of review.
>
> [1] > In the subject, this is just one macro, not multiple macros.

@Dhruva, thanks for the review!

I've fixed up the shortcomings pointed out above and applied the patch
as 6.18 material.

> > ---
> >  drivers/cpufreq/freq_table.c | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c
> > index d5111ee56e38..408fd8fee2e3 100644
> > --- a/drivers/cpufreq/freq_table.c
> > +++ b/drivers/cpufreq/freq_table.c
> > @@ -33,16 +33,17 @@ int cpufreq_frequency_table_cpuinfo(struct cpufreq_policy *policy)
> >       struct cpufreq_frequency_table *pos, *table = policy->freq_table;
> >       unsigned int min_freq = ~0;
> >       unsigned int max_freq = 0;
> > +     unsigned int i;
> >       unsigned int freq;
> >
> > -     cpufreq_for_each_valid_entry(pos, table) {
> > +     cpufreq_for_each_valid_entry_idx(pos, table, i) {
> >               freq = pos->frequency;
> >
> >               if ((!cpufreq_boost_enabled() || !policy->boost_enabled)
> >                   && (pos->flags & CPUFREQ_BOOST_FREQ))
> >                       continue;
> >
> > -             pr_debug("table entry %u: %u kHz\n", (int)(pos - table), freq);
> > +             pr_debug("table entry %u: %u kHz\n", i, freq);
> >               if (freq < min_freq)
> >                       min_freq = freq;
> >               if (freq > max_freq)
> > @@ -126,7 +127,7 @@ int cpufreq_table_index_unsorted(struct cpufreq_policy *policy,
> >       };
> >       struct cpufreq_frequency_table *pos;
> >       struct cpufreq_frequency_table *table = policy->freq_table;
> > -     unsigned int freq, diff, i = 0;
> > +     unsigned int freq, diff, i;
> >       int index;
>
> Usually, it isn't advised to touch code that's not strictly relevant to
> your patch. However since this seems like a minor fixup it's fine by
> me... Upto Rafael whether he prefers to have/not have this unrelated change.
>
> But atleast mention in your commit message that you're also removing the
> initialization from cpufreq_table_index_unsorted as part of some minor cleanup
> (which seems kinda unnecessary to me TBH in the first place)

Yeah, better to say about things like this in the patch changelog.
I've fixed that too.