diff mbox series

[v7,2/6] sched: Add helper kstat_cpu_softirqs_sum()

Message ID 20221111130709.247-3-thunder.leizhen@huawei.com (mailing list archive)
State Superseded
Headers show
Series rcu: Add RCU stall diagnosis information | expand

Commit Message

Leizhen (ThunderTown) Nov. 11, 2022, 1:07 p.m. UTC
Similar to kstat_cpu_irqs_sum(), it counts the sum of all software
interrupts on a specified CPU.

Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
---
 include/linux/kernel_stat.h | 11 +++++++++++
 1 file changed, 11 insertions(+)

Comments

Frederic Weisbecker Nov. 14, 2022, 11:42 a.m. UTC | #1
On Fri, Nov 11, 2022 at 09:07:05PM +0800, Zhen Lei wrote:
> Similar to kstat_cpu_irqs_sum(), it counts the sum of all software
> interrupts on a specified CPU.
> 
> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
> ---
>  include/linux/kernel_stat.h | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
> index 90e2fdc17d79ff8..898076e173a928a 100644
> --- a/include/linux/kernel_stat.h
> +++ b/include/linux/kernel_stat.h
> @@ -67,6 +67,17 @@ static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
>         return kstat_cpu(cpu).softirqs[irq];
>  }
>  
> +static inline unsigned int kstat_cpu_softirqs_sum(int cpu)
> +{
> +	int i;
> +	unsigned int sum = 0;
> +
> +	for (i = 0; i < NR_SOFTIRQS; i++)
> +		sum += kstat_softirqs_cpu(i, cpu);

This should return a u64 or at least an unsigned long so that high
numbers of CPUs don't overflow easily.

Thanks.

> +
> +	return sum;
> +}
> +
>  /*
>   * Number of interrupts per specific IRQ source, since bootup
>   */
> -- 
> 2.25.1
>
Leizhen (ThunderTown) Nov. 14, 2022, 12:45 p.m. UTC | #2
On 2022/11/14 19:42, Frederic Weisbecker wrote:
> On Fri, Nov 11, 2022 at 09:07:05PM +0800, Zhen Lei wrote:
>> Similar to kstat_cpu_irqs_sum(), it counts the sum of all software
>> interrupts on a specified CPU.
>>
>> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
>> ---
>>  include/linux/kernel_stat.h | 11 +++++++++++
>>  1 file changed, 11 insertions(+)
>>
>> diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
>> index 90e2fdc17d79ff8..898076e173a928a 100644
>> --- a/include/linux/kernel_stat.h
>> +++ b/include/linux/kernel_stat.h
>> @@ -67,6 +67,17 @@ static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
>>         return kstat_cpu(cpu).softirqs[irq];
>>  }
>>  
>> +static inline unsigned int kstat_cpu_softirqs_sum(int cpu)
>> +{
>> +	int i;
>> +	unsigned int sum = 0;
>> +
>> +	for (i = 0; i < NR_SOFTIRQS; i++)
>> +		sum += kstat_softirqs_cpu(i, cpu);
> 
> This should return a u64 or at least an unsigned long so that high
> numbers of CPUs don't overflow easily.

OK! Then I'll adjust the following types by the way.

struct kernel_stat {
        unsigned long irqs_sum;
        unsigned int softirqs[NR_SOFTIRQS];
};

static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)




> 
> Thanks.
> 
>> +
>> +	return sum;
>> +}
>> +
>>  /*
>>   * Number of interrupts per specific IRQ source, since bootup
>>   */
>> -- 
>> 2.25.1
>>
> .
>
Frederic Weisbecker Nov. 14, 2022, 12:50 p.m. UTC | #3
On Mon, Nov 14, 2022 at 08:45:26PM +0800, Leizhen (ThunderTown) wrote:
> 
> 
> On 2022/11/14 19:42, Frederic Weisbecker wrote:
> > On Fri, Nov 11, 2022 at 09:07:05PM +0800, Zhen Lei wrote:
> >> Similar to kstat_cpu_irqs_sum(), it counts the sum of all software
> >> interrupts on a specified CPU.
> >>
> >> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
> >> ---
> >>  include/linux/kernel_stat.h | 11 +++++++++++
> >>  1 file changed, 11 insertions(+)
> >>
> >> diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
> >> index 90e2fdc17d79ff8..898076e173a928a 100644
> >> --- a/include/linux/kernel_stat.h
> >> +++ b/include/linux/kernel_stat.h
> >> @@ -67,6 +67,17 @@ static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
> >>         return kstat_cpu(cpu).softirqs[irq];
> >>  }
> >>  
> >> +static inline unsigned int kstat_cpu_softirqs_sum(int cpu)
> >> +{
> >> +	int i;
> >> +	unsigned int sum = 0;
> >> +
> >> +	for (i = 0; i < NR_SOFTIRQS; i++)
> >> +		sum += kstat_softirqs_cpu(i, cpu);
> > 
> > This should return a u64 or at least an unsigned long so that high
> > numbers of CPUs don't overflow easily.
> 
> OK! Then I'll adjust the following types by the way.
> 
> struct kernel_stat {
>         unsigned long irqs_sum;
>         unsigned int softirqs[NR_SOFTIRQS];
> };
> 
> static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)

And I also misread kstat_softirqs_cpu() content, my bad. Although at some point
the sum of all softirqs within a single CPU might overflow in an int so I guess
that still applies...

Thanks.


> 
> 
> 
> 
> > 
> > Thanks.
> > 
> >> +
> >> +	return sum;
> >> +}
> >> +
> >>  /*
> >>   * Number of interrupts per specific IRQ source, since bootup
> >>   */
> >> -- 
> >> 2.25.1
> >>
> > .
> > 
> 
> -- 
> Regards,
>   Zhen Lei
Leizhen (ThunderTown) Nov. 14, 2022, 2:26 p.m. UTC | #4
On 2022/11/14 20:50, Frederic Weisbecker wrote:
> On Mon, Nov 14, 2022 at 08:45:26PM +0800, Leizhen (ThunderTown) wrote:
>>
>>
>> On 2022/11/14 19:42, Frederic Weisbecker wrote:
>>> On Fri, Nov 11, 2022 at 09:07:05PM +0800, Zhen Lei wrote:
>>>> Similar to kstat_cpu_irqs_sum(), it counts the sum of all software
>>>> interrupts on a specified CPU.
>>>>
>>>> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
>>>> ---
>>>>  include/linux/kernel_stat.h | 11 +++++++++++
>>>>  1 file changed, 11 insertions(+)
>>>>
>>>> diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
>>>> index 90e2fdc17d79ff8..898076e173a928a 100644
>>>> --- a/include/linux/kernel_stat.h
>>>> +++ b/include/linux/kernel_stat.h
>>>> @@ -67,6 +67,17 @@ static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
>>>>         return kstat_cpu(cpu).softirqs[irq];
>>>>  }
>>>>  
>>>> +static inline unsigned int kstat_cpu_softirqs_sum(int cpu)
>>>> +{
>>>> +	int i;
>>>> +	unsigned int sum = 0;
>>>> +
>>>> +	for (i = 0; i < NR_SOFTIRQS; i++)
>>>> +		sum += kstat_softirqs_cpu(i, cpu);
>>>
>>> This should return a u64 or at least an unsigned long so that high
>>> numbers of CPUs don't overflow easily.
>>
>> OK! Then I'll adjust the following types by the way.
>>
>> struct kernel_stat {
>>         unsigned long irqs_sum;
>>         unsigned int softirqs[NR_SOFTIRQS];
>> };
>>
>> static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
> 
> And I also misread kstat_softirqs_cpu() content, my bad. Although at some point
> the sum of all softirqs within a single CPU might overflow in an int so I guess
> that still applies...

Yes, this was discussed before. In fact, I think unsigned int and unsigned
long are all right.

> 
> Thanks.
> 
> 
>>
>>
>>
>>
>>>
>>> Thanks.
>>>
>>>> +
>>>> +	return sum;
>>>> +}
>>>> +
>>>>  /*
>>>>   * Number of interrupts per specific IRQ source, since bootup
>>>>   */
>>>> -- 
>>>> 2.25.1
>>>>
>>> .
>>>
>>
>> -- 
>> Regards,
>>   Zhen Lei
> .
>
diff mbox series

Patch

diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
index 90e2fdc17d79ff8..898076e173a928a 100644
--- a/include/linux/kernel_stat.h
+++ b/include/linux/kernel_stat.h
@@ -67,6 +67,17 @@  static inline unsigned int kstat_softirqs_cpu(unsigned int irq, int cpu)
        return kstat_cpu(cpu).softirqs[irq];
 }
 
+static inline unsigned int kstat_cpu_softirqs_sum(int cpu)
+{
+	int i;
+	unsigned int sum = 0;
+
+	for (i = 0; i < NR_SOFTIRQS; i++)
+		sum += kstat_softirqs_cpu(i, cpu);
+
+	return sum;
+}
+
 /*
  * Number of interrupts per specific IRQ source, since bootup
  */