[PATCH] arch_topology: init capacity_freq_ref to 0

[PATCH] arch_topology: init capacity_freq_ref to 0

[Ionela Voinescu]

It's useful to have capacity_freq_ref initialized to 0 for users of
arch_scale_freq_ref() to detect when capacity_freq_ref was not
yet set.

The only scenario affected by this change in the init value is when a
cpufreq driver is never loaded. As a result, the only setter of a
cpu scale factor remains the call of topology_normalize_cpu_scale()
from parse_dt_topology(). There we cannot use the value 0 of
capacity_freq_ref so we have to compensate for its uninitialized state.

Signed-off-by: Ionela Voinescu <ionela.voinescu@arm.com>
Signed-off-by: Beata Michalska <beata.michalska@arm.com>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
---

Hi,

This patch was previously part of the patch-set at [1], but we thought
it's best to separate the standalone patches in that set to make
further review easier on the remaining topics.

Based on v6.11-rc5 and tested on Juno with and without a cpufreq driver.

[1] https://lore.kernel.org/lkml/20240603082154.3830591-2-beata.michalska@arm.com/

Thanks,
Ionela.

 drivers/base/arch_topology.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
index 75fcb75d5515..c49ef1a712f4 100644
--- a/drivers/base/arch_topology.c
+++ b/drivers/base/arch_topology.c
@@ -28,7 +28,7 @@
 static DEFINE_PER_CPU(struct scale_freq_data __rcu *, sft_data);
 static struct cpumask scale_freq_counters_mask;
 static bool scale_freq_invariant;
-DEFINE_PER_CPU(unsigned long, capacity_freq_ref) = 1;
+DEFINE_PER_CPU(unsigned long, capacity_freq_ref) = 0;
 EXPORT_PER_CPU_SYMBOL_GPL(capacity_freq_ref);
 
 static bool supports_scale_freq_counters(const struct cpumask *cpus)
@@ -293,13 +293,15 @@ void topology_normalize_cpu_scale(void)
 
 	capacity_scale = 1;
 	for_each_possible_cpu(cpu) {
-		capacity = raw_capacity[cpu] * per_cpu(capacity_freq_ref, cpu);
+		capacity = raw_capacity[cpu] *
+			   (per_cpu(capacity_freq_ref, cpu) ?: 1);
 		capacity_scale = max(capacity, capacity_scale);
 	}
 
 	pr_debug("cpu_capacity: capacity_scale=%llu\n", capacity_scale);
 	for_each_possible_cpu(cpu) {
-		capacity = raw_capacity[cpu] * per_cpu(capacity_freq_ref, cpu);
+		capacity = raw_capacity[cpu] *
+			   (per_cpu(capacity_freq_ref, cpu) ?: 1);
 		capacity = div64_u64(capacity << SCHED_CAPACITY_SHIFT,
 			capacity_scale);
 		topology_set_cpu_scale(cpu, capacity);
-- 
2.25.1

Re: [PATCH] arch_topology: init capacity_freq_ref to 0

[Ionela Voinescu]

Hi folks,

I just wanted to mention that this patch still applies cleanly on
next-20250217 as well, and it still builds/boots/works as expected.

I've rechecked it given that the patches at [1] seem ready to be picked
up, and those patches depend on this one.

Thanks,
Ionela.

[1] https://lore.kernel.org/lkml/20250131162439.3843071-1-beata.michalska@arm.com/

On Tuesday 27 Aug 2024 at 16:48:18 (+0100), Ionela Voinescu wrote:
> It's useful to have capacity_freq_ref initialized to 0 for users of
> arch_scale_freq_ref() to detect when capacity_freq_ref was not
> yet set.
> 
> The only scenario affected by this change in the init value is when a
> cpufreq driver is never loaded. As a result, the only setter of a
> cpu scale factor remains the call of topology_normalize_cpu_scale()
> from parse_dt_topology(). There we cannot use the value 0 of
> capacity_freq_ref so we have to compensate for its uninitialized state.
> 
> Signed-off-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Signed-off-by: Beata Michalska <beata.michalska@arm.com>
> Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
> Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
> ---
> 
> Hi,
> 
> This patch was previously part of the patch-set at [1], but we thought
> it's best to separate the standalone patches in that set to make
> further review easier on the remaining topics.
> 
> Based on v6.11-rc5 and tested on Juno with and without a cpufreq driver.
> 
> [1] https://lore.kernel.org/lkml/20240603082154.3830591-2-beata.michalska@arm.com/
> 
> Thanks,
> Ionela.
> 
>  drivers/base/arch_topology.c | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
> index 75fcb75d5515..c49ef1a712f4 100644
> --- a/drivers/base/arch_topology.c
> +++ b/drivers/base/arch_topology.c
> @@ -28,7 +28,7 @@
>  static DEFINE_PER_CPU(struct scale_freq_data __rcu *, sft_data);
>  static struct cpumask scale_freq_counters_mask;
>  static bool scale_freq_invariant;
> -DEFINE_PER_CPU(unsigned long, capacity_freq_ref) = 1;
> +DEFINE_PER_CPU(unsigned long, capacity_freq_ref) = 0;
>  EXPORT_PER_CPU_SYMBOL_GPL(capacity_freq_ref);
>  
>  static bool supports_scale_freq_counters(const struct cpumask *cpus)
> @@ -293,13 +293,15 @@ void topology_normalize_cpu_scale(void)
>  
>  	capacity_scale = 1;
>  	for_each_possible_cpu(cpu) {
> -		capacity = raw_capacity[cpu] * per_cpu(capacity_freq_ref, cpu);
> +		capacity = raw_capacity[cpu] *
> +			   (per_cpu(capacity_freq_ref, cpu) ?: 1);
>  		capacity_scale = max(capacity, capacity_scale);
>  	}
>  
>  	pr_debug("cpu_capacity: capacity_scale=%llu\n", capacity_scale);
>  	for_each_possible_cpu(cpu) {
> -		capacity = raw_capacity[cpu] * per_cpu(capacity_freq_ref, cpu);
> +		capacity = raw_capacity[cpu] *
> +			   (per_cpu(capacity_freq_ref, cpu) ?: 1);
>  		capacity = div64_u64(capacity << SCHED_CAPACITY_SHIFT,
>  			capacity_scale);
>  		topology_set_cpu_scale(cpu, capacity);
> -- 
> 2.25.1
> 
> 

Re: [PATCH] arch_topology: init capacity_freq_ref to 0

[Sudeep Holla]

On Mon, Feb 17, 2025 at 03:01:23PM +0000, Ionela Voinescu wrote:
> Hi folks,
> 
> I just wanted to mention that this patch still applies cleanly on
> next-20250217 as well, and it still builds/boots/works as expected.
> 
> I've rechecked it given that the patches at [1] seem ready to be picked
> up, and those patches depend on this one.
>

Is there a dependency on [1] ?

Either way, if Catalin is happy to pick it up along with [1], that would
be best and easiest.

If not and there are not dependencies, then we can ask Greg to pick up.
Which means you need to repost with him cc-ed ? Lets wait and see if
Catalin is happy to pick up before reposting though.

--
Regards,
Sudeep

Re: [PATCH] arch_topology: init capacity_freq_ref to 0

[Catalin Marinas]

On Tue, 27 Aug 2024 16:48:18 +0100, Ionela Voinescu wrote:
> It's useful to have capacity_freq_ref initialized to 0 for users of
> arch_scale_freq_ref() to detect when capacity_freq_ref was not
> yet set.
> 
> The only scenario affected by this change in the init value is when a
> cpufreq driver is never loaded. As a result, the only setter of a
> cpu scale factor remains the call of topology_normalize_cpu_scale()
> from parse_dt_topology(). There we cannot use the value 0 of
> capacity_freq_ref so we have to compensate for its uninitialized state.
> 
> [...]

Applied to arm64 (for-next/amuv1-avg-freq), thanks!

[1/1] arch_topology: init capacity_freq_ref to 0
      https://git.kernel.org/arm64/c/004b500a9031

-- 
Catalin