In the worst case, the freq_table of policy data is not sorted and
contains duplicate frequencies, this means that it needs to iterate
through the entire freq_table of policy to ensure each frequency is
unique in the freq_table of stats data, this has a time complexity of
O(N^2), where N is the number of frequencies in the freq_table of
policy.

However, if the policy.freq_table is already sorted and contains no
duplicate frequencices, it can reduce the time complexity of creating
stats.freq_table to O(N), the 'freq_table_sorted' field of policy data
can be used to indicate whether the policy.freq_table is sorted.

Signed-off-by: Liao Chang <liaochang1@huawei.com>
---
 drivers/cpufreq/cpufreq_stats.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
index 55c7ffd37d1c..fcb74050711a 100644
--- a/drivers/cpufreq/cpufreq_stats.c
+++ b/drivers/cpufreq/cpufreq_stats.c
@@ -243,7 +243,8 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
 
 	/* Find valid-unique entries */
 	cpufreq_for_each_valid_entry(pos, policy->freq_table)
-		if (freq_table_get_index(stats, pos->frequency) == -1)
+		if ((policy->freq_table_sorted != CPUFREQ_TABLE_UNSORTED) ||
+		    (freq_table_get_index(stats, pos->frequency) == -1))
 			stats->freq_table[i++] = pos->frequency;
 
 	stats->state_num = i;
-- 
2.34.1

On Aug 18, 2023 at 09:50:00 +0000, Liao Chang wrote:
> In the worst case, the freq_table of policy data is not sorted and
> contains duplicate frequencies, this means that it needs to iterate
> through the entire freq_table of policy to ensure each frequency is
> unique in the freq_table of stats data, this has a time complexity of
> O(N^2), where N is the number of frequencies in the freq_table of
> policy.
> 
> However, if the policy.freq_table is already sorted and contains no
> duplicate frequencices, it can reduce the time complexity of creating

s/frequencices/frequencies?

> stats.freq_table to O(N), the 'freq_table_sorted' field of policy data
> can be used to indicate whether the policy.freq_table is sorted.
> 
> Signed-off-by: Liao Chang <liaochang1@huawei.com>
> ---
>  drivers/cpufreq/cpufreq_stats.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
> index 55c7ffd37d1c..fcb74050711a 100644
> --- a/drivers/cpufreq/cpufreq_stats.c
> +++ b/drivers/cpufreq/cpufreq_stats.c
> @@ -243,7 +243,8 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
>  
>  	/* Find valid-unique entries */
>  	cpufreq_for_each_valid_entry(pos, policy->freq_table)
> -		if (freq_table_get_index(stats, pos->frequency) == -1)
> +		if ((policy->freq_table_sorted != CPUFREQ_TABLE_UNSORTED) ||

[...]

Otherwise looks okay to me,

Reviewed-by: Dhruva Gole <d-gole@ti.com>

-- 
Best regards,
Dhruva Gole <d-gole@ti.com>

On Mon, Aug 21, 2023 at 11:14 AM Dhruva Gole <d-gole@ti.com> wrote:
>
> On Aug 18, 2023 at 09:50:00 +0000, Liao Chang wrote:
> > In the worst case, the freq_table of policy data is not sorted and
> > contains duplicate frequencies, this means that it needs to iterate
> > through the entire freq_table of policy to ensure each frequency is
> > unique in the freq_table of stats data, this has a time complexity of
> > O(N^2), where N is the number of frequencies in the freq_table of
> > policy.
> >
> > However, if the policy.freq_table is already sorted and contains no
> > duplicate frequencices, it can reduce the time complexity of creating
>
> s/frequencices/frequencies?

I've fixed this when applying the patch.

> > stats.freq_table to O(N), the 'freq_table_sorted' field of policy data
> > can be used to indicate whether the policy.freq_table is sorted.
> >
> > Signed-off-by: Liao Chang <liaochang1@huawei.com>
> > ---
> >  drivers/cpufreq/cpufreq_stats.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
> > index 55c7ffd37d1c..fcb74050711a 100644
> > --- a/drivers/cpufreq/cpufreq_stats.c
> > +++ b/drivers/cpufreq/cpufreq_stats.c
> > @@ -243,7 +243,8 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
> >
> >       /* Find valid-unique entries */
> >       cpufreq_for_each_valid_entry(pos, policy->freq_table)
> > -             if (freq_table_get_index(stats, pos->frequency) == -1)
> > +             if ((policy->freq_table_sorted != CPUFREQ_TABLE_UNSORTED) ||

I've also removed the redundant parens from this check.

>
> [...]
>
> Otherwise looks okay to me,
>
> Reviewed-by: Dhruva Gole <d-gole@ti.com>

Applied as 6.6 material (with the changes mentioned above), thanks!

On 18-08-23, 09:50, Liao Chang wrote:
> In the worst case, the freq_table of policy data is not sorted and
> contains duplicate frequencies, this means that it needs to iterate
> through the entire freq_table of policy to ensure each frequency is
> unique in the freq_table of stats data, this has a time complexity of
> O(N^2), where N is the number of frequencies in the freq_table of
> policy.
> 
> However, if the policy.freq_table is already sorted and contains no
> duplicate frequencices, it can reduce the time complexity of creating
> stats.freq_table to O(N), the 'freq_table_sorted' field of policy data
> can be used to indicate whether the policy.freq_table is sorted.
> 
> Signed-off-by: Liao Chang <liaochang1@huawei.com>
> ---
>  drivers/cpufreq/cpufreq_stats.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
> index 55c7ffd37d1c..fcb74050711a 100644
> --- a/drivers/cpufreq/cpufreq_stats.c
> +++ b/drivers/cpufreq/cpufreq_stats.c
> @@ -243,7 +243,8 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
>  
>  	/* Find valid-unique entries */
>  	cpufreq_for_each_valid_entry(pos, policy->freq_table)
> -		if (freq_table_get_index(stats, pos->frequency) == -1)
> +		if ((policy->freq_table_sorted != CPUFREQ_TABLE_UNSORTED) ||
> +		    (freq_table_get_index(stats, pos->frequency) == -1))
>  			stats->freq_table[i++] = pos->frequency;
>  
>  	stats->state_num = i;

Acked-by: Viresh Kumar <viresh.kumar@linaro.org>

-- 
viresh