diff mbox series

[v4,02/11] perf/x86/rapl: Fix the energy-pkg event for AMD CPUs

Message ID 20240711102436.4432-3-Dhananjay.Ugwekar@amd.com (mailing list archive)
State Handled Elsewhere, archived
Headers show
Series Add per-core RAPL energy counter support for AMD CPUs | expand

Commit Message

Dhananjay Ugwekar July 11, 2024, 10:24 a.m. UTC
After commit ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf"),
on AMD processors that support extended CPUID leaf 0x80000026, the
topology_die_cpumask() and topology_logical_die_id() macros, no longer
return the package cpumask and package id, instead they return the CCD
(Core Complex Die) mask and id respectively. This leads to the energy-pkg
event scope to be modified to CCD instead of package.

Replacing these macros with their package counterparts fixes the
energy-pkg event for AMD CPUs.

However due to the difference between the scope of energy-pkg event for
Intel and AMD CPUs, we have to replace these macros conditionally only for
AMD CPUs.

On a 12 CCD 1 Package AMD Zen4 Genoa machine:

Before:
$ cat /sys/devices/power/cpumask
0,8,16,24,32,40,48,56,64,72,80,88.

The expected cpumask here is supposed to be just "0", as it is a package
scope event, only one CPU will be collecting the event for all the CPUs in
the package.

After:
$ cat /sys/devices/power/cpumask
0

Signed-off-by: Dhananjay Ugwekar <Dhananjay.Ugwekar@amd.com>
Fixes: 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf")
---
Changes in v4:
* Invert the pkg scope check in init_rapl_pmus() (Peter)
* Add comments to explain the pkg scope check (Peter)

PS: Scope check logic is still kept the same (i.e., all Intel systems being 
considered as die scope), Rui will be modifying it to limit the die-scope 
only to Cascadelake-AP in a future patch on top of this patchset.
---
 arch/x86/events/rapl.c | 39 ++++++++++++++++++++++++++++++++++-----
 1 file changed, 34 insertions(+), 5 deletions(-)

Comments

Zhang, Rui July 12, 2024, 2:04 a.m. UTC | #1
On Thu, 2024-07-11 at 10:24 +0000, Dhananjay Ugwekar wrote:
> After commit ("x86/cpu/topology: Add support for the AMD 0x80000026
> leaf"),
> on AMD processors that support extended CPUID leaf 0x80000026, the
> topology_die_cpumask() and topology_logical_die_id() macros, no
> longer
> return the package cpumask and package id, instead they return the
> CCD
> (Core Complex Die) mask and id respectively. This leads to the
> energy-pkg
> event scope to be modified to CCD instead of package.
> 
> Replacing these macros with their package counterparts fixes the
> energy-pkg event for AMD CPUs.
> 
> However due to the difference between the scope of energy-pkg event
> for
> Intel and AMD CPUs, we have to replace these macros conditionally
> only for
> AMD CPUs.
> 
> On a 12 CCD 1 Package AMD Zen4 Genoa machine:
> 
> Before:
> $ cat /sys/devices/power/cpumask
> 0,8,16,24,32,40,48,56,64,72,80,88.
> 
> The expected cpumask here is supposed to be just "0", as it is a
> package
> scope event, only one CPU will be collecting the event for all the
> CPUs in
> the package.
> 
> After:
> $ cat /sys/devices/power/cpumask
> 0
> 
> Signed-off-by: Dhananjay Ugwekar <Dhananjay.Ugwekar@amd.com>
> Fixes: 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD
> 0x80000026 leaf")

You still missed my Reviewed-by tag for this one. :)
https://lore.kernel.org/all/e1f70a09f85dbd0ee3f32dffea37993e141269d0.camel@intel.com/

thanks,
rui

> ---
> Changes in v4:
> * Invert the pkg scope check in init_rapl_pmus() (Peter)
> * Add comments to explain the pkg scope check (Peter)
> 
> PS: Scope check logic is still kept the same (i.e., all Intel systems
> being 
> considered as die scope), Rui will be modifying it to limit the die-
> scope 
> only to Cascadelake-AP in a future patch on top of this patchset.
> ---
>  arch/x86/events/rapl.c | 39 ++++++++++++++++++++++++++++++++++-----
>  1 file changed, 34 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
> index 0c5e7a7c43ac..df71f38ad98d 100644
> --- a/arch/x86/events/rapl.c
> +++ b/arch/x86/events/rapl.c
> @@ -103,6 +103,13 @@ static struct perf_pmu_events_attr
> event_attr_##v = {                              \
>         .event_str      =
> str,                                                  \
>  };
>  
> +/*
> + * RAPL PMU scope for AMD is package whereas for Intel it is die.
> + */
> +#define rapl_pmu_is_pkg_scope()                                \
> +       (boot_cpu_data.x86_vendor == X86_VENDOR_AMD ||  \
> +        boot_cpu_data.x86_vendor == X86_VENDOR_HYGON)
> +
>  struct rapl_pmu {
>         raw_spinlock_t          lock;
>         int                     n_active;
> @@ -140,9 +147,25 @@ static unsigned int rapl_cntr_mask;
>  static u64 rapl_timer_ms;
>  static struct perf_msr *rapl_msrs;
>  
> +/*
> + * Helper functions to get the correct topology macros according to
> the
> + * RAPL PMU scope.
> + */
> +static inline unsigned int get_rapl_pmu_idx(int cpu)
> +{
> +       return rapl_pmu_is_pkg_scope() ?
> topology_logical_package_id(cpu) :
> +                                       
> topology_logical_die_id(cpu);
> +}
> +
> +static inline const struct cpumask *get_rapl_pmu_cpumask(int cpu)
> +{
> +       return rapl_pmu_is_pkg_scope() ? topology_core_cpumask(cpu) :
> +                                        topology_die_cpumask(cpu);
> +}
> +
>  static inline struct rapl_pmu *cpu_to_rapl_pmu(unsigned int cpu)
>  {
> -       unsigned int rapl_pmu_idx = topology_logical_die_id(cpu);
> +       unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
>  
>         /*
>          * The unsigned check also catches the '-1' return value for
> non
> @@ -543,6 +566,7 @@ static struct perf_msr amd_rapl_msrs[] = {
>  
>  static int rapl_cpu_offline(unsigned int cpu)
>  {
> +       const struct cpumask *rapl_pmu_cpumask =
> get_rapl_pmu_cpumask(cpu);
>         struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>         int target;
>  
> @@ -552,7 +576,7 @@ static int rapl_cpu_offline(unsigned int cpu)
>  
>         pmu->cpu = -1;
>         /* Find a new cpu to collect rapl events */
> -       target = cpumask_any_but(topology_die_cpumask(cpu), cpu);
> +       target = cpumask_any_but(rapl_pmu_cpumask, cpu);
>  
>         /* Migrate rapl events to the new target */
>         if (target < nr_cpu_ids) {
> @@ -565,6 +589,8 @@ static int rapl_cpu_offline(unsigned int cpu)
>  
>  static int rapl_cpu_online(unsigned int cpu)
>  {
> +       unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
> +       const struct cpumask *rapl_pmu_cpumask =
> get_rapl_pmu_cpumask(cpu);
>         struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>         int target;
>  
> @@ -579,14 +605,14 @@ static int rapl_cpu_online(unsigned int cpu)
>                 pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
>                 rapl_hrtimer_init(pmu);
>  
> -               rapl_pmus->pmus[topology_logical_die_id(cpu)] = pmu;
> +               rapl_pmus->pmus[rapl_pmu_idx] = pmu;
>         }
>  
>         /*
>          * Check if there is an online cpu in the package which
> collects rapl
>          * events already.
>          */
> -       target = cpumask_any_and(&rapl_cpu_mask,
> topology_die_cpumask(cpu));
> +       target = cpumask_any_and(&rapl_cpu_mask, rapl_pmu_cpumask);
>         if (target < nr_cpu_ids)
>                 return 0;
>  
> @@ -675,7 +701,10 @@ static const struct attribute_group
> *rapl_attr_update[] = {
>  
>  static int __init init_rapl_pmus(void)
>  {
> -       int nr_rapl_pmu = topology_max_packages() *
> topology_max_dies_per_package();
> +       int nr_rapl_pmu = topology_max_packages();
> +
> +       if (!rapl_pmu_is_pkg_scope())
> +               nr_rapl_pmu *= topology_max_dies_per_package();
>  
>         rapl_pmus = kzalloc(struct_size(rapl_pmus, pmus,
> nr_rapl_pmu), GFP_KERNEL);
>         if (!rapl_pmus)
Dhananjay Ugwekar July 12, 2024, 3:28 a.m. UTC | #2
Hello Rui,

On 7/12/2024 7:34 AM, Zhang, Rui wrote:
> On Thu, 2024-07-11 at 10:24 +0000, Dhananjay Ugwekar wrote:
>> After commit ("x86/cpu/topology: Add support for the AMD 0x80000026
>> leaf"),
>> on AMD processors that support extended CPUID leaf 0x80000026, the
>> topology_die_cpumask() and topology_logical_die_id() macros, no
>> longer
>> return the package cpumask and package id, instead they return the
>> CCD
>> (Core Complex Die) mask and id respectively. This leads to the
>> energy-pkg
>> event scope to be modified to CCD instead of package.
>>
>> Replacing these macros with their package counterparts fixes the
>> energy-pkg event for AMD CPUs.
>>
>> However due to the difference between the scope of energy-pkg event
>> for
>> Intel and AMD CPUs, we have to replace these macros conditionally
>> only for
>> AMD CPUs.
>>
>> On a 12 CCD 1 Package AMD Zen4 Genoa machine:
>>
>> Before:
>> $ cat /sys/devices/power/cpumask
>> 0,8,16,24,32,40,48,56,64,72,80,88.
>>
>> The expected cpumask here is supposed to be just "0", as it is a
>> package
>> scope event, only one CPU will be collecting the event for all the
>> CPUs in
>> the package.
>>
>> After:
>> $ cat /sys/devices/power/cpumask
>> 0
>>
>> Signed-off-by: Dhananjay Ugwekar <Dhananjay.Ugwekar@amd.com>
>> Fixes: 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD
>> 0x80000026 leaf")
> 
> You still missed my Reviewed-by tag for this one. :)
> https://lore.kernel.org/all/e1f70a09f85dbd0ee3f32dffea37993e141269d0.camel@intel.com/

I didnt forget this time, just wanted to give you a chance to review the changes :) (although they are minimal)

Will surely add it in next version! 

Thanks,
Dhananjay

> 
> thanks,
> rui
> 
>> ---
>> Changes in v4:
>> * Invert the pkg scope check in init_rapl_pmus() (Peter)
>> * Add comments to explain the pkg scope check (Peter)
>>
>> PS: Scope check logic is still kept the same (i.e., all Intel systems
>> being 
>> considered as die scope), Rui will be modifying it to limit the die-
>> scope 
>> only to Cascadelake-AP in a future patch on top of this patchset.
>> ---
>>  arch/x86/events/rapl.c | 39 ++++++++++++++++++++++++++++++++++-----
>>  1 file changed, 34 insertions(+), 5 deletions(-)
>>
>> diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
>> index 0c5e7a7c43ac..df71f38ad98d 100644
>> --- a/arch/x86/events/rapl.c
>> +++ b/arch/x86/events/rapl.c
>> @@ -103,6 +103,13 @@ static struct perf_pmu_events_attr
>> event_attr_##v = {                              \
>>         .event_str      =
>> str,                                                  \
>>  };
>>  
>> +/*
>> + * RAPL PMU scope for AMD is package whereas for Intel it is die.
>> + */
>> +#define rapl_pmu_is_pkg_scope()                                \
>> +       (boot_cpu_data.x86_vendor == X86_VENDOR_AMD ||  \
>> +        boot_cpu_data.x86_vendor == X86_VENDOR_HYGON)
>> +
>>  struct rapl_pmu {
>>         raw_spinlock_t          lock;
>>         int                     n_active;
>> @@ -140,9 +147,25 @@ static unsigned int rapl_cntr_mask;
>>  static u64 rapl_timer_ms;
>>  static struct perf_msr *rapl_msrs;
>>  
>> +/*
>> + * Helper functions to get the correct topology macros according to
>> the
>> + * RAPL PMU scope.
>> + */
>> +static inline unsigned int get_rapl_pmu_idx(int cpu)
>> +{
>> +       return rapl_pmu_is_pkg_scope() ?
>> topology_logical_package_id(cpu) :
>> +                                       
>> topology_logical_die_id(cpu);
>> +}
>> +
>> +static inline const struct cpumask *get_rapl_pmu_cpumask(int cpu)
>> +{
>> +       return rapl_pmu_is_pkg_scope() ? topology_core_cpumask(cpu) :
>> +                                        topology_die_cpumask(cpu);
>> +}
>> +
>>  static inline struct rapl_pmu *cpu_to_rapl_pmu(unsigned int cpu)
>>  {
>> -       unsigned int rapl_pmu_idx = topology_logical_die_id(cpu);
>> +       unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
>>  
>>         /*
>>          * The unsigned check also catches the '-1' return value for
>> non
>> @@ -543,6 +566,7 @@ static struct perf_msr amd_rapl_msrs[] = {
>>  
>>  static int rapl_cpu_offline(unsigned int cpu)
>>  {
>> +       const struct cpumask *rapl_pmu_cpumask =
>> get_rapl_pmu_cpumask(cpu);
>>         struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>>         int target;
>>  
>> @@ -552,7 +576,7 @@ static int rapl_cpu_offline(unsigned int cpu)
>>  
>>         pmu->cpu = -1;
>>         /* Find a new cpu to collect rapl events */
>> -       target = cpumask_any_but(topology_die_cpumask(cpu), cpu);
>> +       target = cpumask_any_but(rapl_pmu_cpumask, cpu);
>>  
>>         /* Migrate rapl events to the new target */
>>         if (target < nr_cpu_ids) {
>> @@ -565,6 +589,8 @@ static int rapl_cpu_offline(unsigned int cpu)
>>  
>>  static int rapl_cpu_online(unsigned int cpu)
>>  {
>> +       unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
>> +       const struct cpumask *rapl_pmu_cpumask =
>> get_rapl_pmu_cpumask(cpu);
>>         struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>>         int target;
>>  
>> @@ -579,14 +605,14 @@ static int rapl_cpu_online(unsigned int cpu)
>>                 pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
>>                 rapl_hrtimer_init(pmu);
>>  
>> -               rapl_pmus->pmus[topology_logical_die_id(cpu)] = pmu;
>> +               rapl_pmus->pmus[rapl_pmu_idx] = pmu;
>>         }
>>  
>>         /*
>>          * Check if there is an online cpu in the package which
>> collects rapl
>>          * events already.
>>          */
>> -       target = cpumask_any_and(&rapl_cpu_mask,
>> topology_die_cpumask(cpu));
>> +       target = cpumask_any_and(&rapl_cpu_mask, rapl_pmu_cpumask);
>>         if (target < nr_cpu_ids)
>>                 return 0;
>>  
>> @@ -675,7 +701,10 @@ static const struct attribute_group
>> *rapl_attr_update[] = {
>>  
>>  static int __init init_rapl_pmus(void)
>>  {
>> -       int nr_rapl_pmu = topology_max_packages() *
>> topology_max_dies_per_package();
>> +       int nr_rapl_pmu = topology_max_packages();
>> +
>> +       if (!rapl_pmu_is_pkg_scope())
>> +               nr_rapl_pmu *= topology_max_dies_per_package();
>>  
>>         rapl_pmus = kzalloc(struct_size(rapl_pmus, pmus,
>> nr_rapl_pmu), GFP_KERNEL);
>>         if (!rapl_pmus)
>
Dhananjay Ugwekar July 16, 2024, 8:56 a.m. UTC | #3
Hello Peter, Rui,

After Ian's comments on the series, I have decided to rethink the approach of
adding a new PMU for the per-core RAPL counters.

However this patch is still needed as a fix for the 
commit ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf"), I will be
sending this separately along with a similar fix for powercap/intel_rapl_common.

Thanks,
Dhananjay

On 7/11/2024 3:54 PM, Dhananjay Ugwekar wrote:
> After commit ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf"),
> on AMD processors that support extended CPUID leaf 0x80000026, the
> topology_die_cpumask() and topology_logical_die_id() macros, no longer
> return the package cpumask and package id, instead they return the CCD
> (Core Complex Die) mask and id respectively. This leads to the energy-pkg
> event scope to be modified to CCD instead of package.
> 
> Replacing these macros with their package counterparts fixes the
> energy-pkg event for AMD CPUs.
> 
> However due to the difference between the scope of energy-pkg event for
> Intel and AMD CPUs, we have to replace these macros conditionally only for
> AMD CPUs.
> 
> On a 12 CCD 1 Package AMD Zen4 Genoa machine:
> 
> Before:
> $ cat /sys/devices/power/cpumask
> 0,8,16,24,32,40,48,56,64,72,80,88.
> 
> The expected cpumask here is supposed to be just "0", as it is a package
> scope event, only one CPU will be collecting the event for all the CPUs in
> the package.
> 
> After:
> $ cat /sys/devices/power/cpumask
> 0
> 
> Signed-off-by: Dhananjay Ugwekar <Dhananjay.Ugwekar@amd.com>
> Fixes: 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf")
> ---
> Changes in v4:
> * Invert the pkg scope check in init_rapl_pmus() (Peter)
> * Add comments to explain the pkg scope check (Peter)
> 
> PS: Scope check logic is still kept the same (i.e., all Intel systems being 
> considered as die scope), Rui will be modifying it to limit the die-scope 
> only to Cascadelake-AP in a future patch on top of this patchset.
> ---
>  arch/x86/events/rapl.c | 39 ++++++++++++++++++++++++++++++++++-----
>  1 file changed, 34 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
> index 0c5e7a7c43ac..df71f38ad98d 100644
> --- a/arch/x86/events/rapl.c
> +++ b/arch/x86/events/rapl.c
> @@ -103,6 +103,13 @@ static struct perf_pmu_events_attr event_attr_##v = {				\
>  	.event_str	= str,							\
>  };
>  
> +/*
> + * RAPL PMU scope for AMD is package whereas for Intel it is die.
> + */
> +#define rapl_pmu_is_pkg_scope()				\
> +	(boot_cpu_data.x86_vendor == X86_VENDOR_AMD ||	\
> +	 boot_cpu_data.x86_vendor == X86_VENDOR_HYGON)
> +
>  struct rapl_pmu {
>  	raw_spinlock_t		lock;
>  	int			n_active;
> @@ -140,9 +147,25 @@ static unsigned int rapl_cntr_mask;
>  static u64 rapl_timer_ms;
>  static struct perf_msr *rapl_msrs;
>  
> +/*
> + * Helper functions to get the correct topology macros according to the
> + * RAPL PMU scope.
> + */
> +static inline unsigned int get_rapl_pmu_idx(int cpu)
> +{
> +	return rapl_pmu_is_pkg_scope() ? topology_logical_package_id(cpu) :
> +					 topology_logical_die_id(cpu);
> +}
> +
> +static inline const struct cpumask *get_rapl_pmu_cpumask(int cpu)
> +{
> +	return rapl_pmu_is_pkg_scope() ? topology_core_cpumask(cpu) :
> +					 topology_die_cpumask(cpu);
> +}
> +
>  static inline struct rapl_pmu *cpu_to_rapl_pmu(unsigned int cpu)
>  {
> -	unsigned int rapl_pmu_idx = topology_logical_die_id(cpu);
> +	unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
>  
>  	/*
>  	 * The unsigned check also catches the '-1' return value for non
> @@ -543,6 +566,7 @@ static struct perf_msr amd_rapl_msrs[] = {
>  
>  static int rapl_cpu_offline(unsigned int cpu)
>  {
> +	const struct cpumask *rapl_pmu_cpumask = get_rapl_pmu_cpumask(cpu);
>  	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>  	int target;
>  
> @@ -552,7 +576,7 @@ static int rapl_cpu_offline(unsigned int cpu)
>  
>  	pmu->cpu = -1;
>  	/* Find a new cpu to collect rapl events */
> -	target = cpumask_any_but(topology_die_cpumask(cpu), cpu);
> +	target = cpumask_any_but(rapl_pmu_cpumask, cpu);
>  
>  	/* Migrate rapl events to the new target */
>  	if (target < nr_cpu_ids) {
> @@ -565,6 +589,8 @@ static int rapl_cpu_offline(unsigned int cpu)
>  
>  static int rapl_cpu_online(unsigned int cpu)
>  {
> +	unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
> +	const struct cpumask *rapl_pmu_cpumask = get_rapl_pmu_cpumask(cpu);
>  	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
>  	int target;
>  
> @@ -579,14 +605,14 @@ static int rapl_cpu_online(unsigned int cpu)
>  		pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
>  		rapl_hrtimer_init(pmu);
>  
> -		rapl_pmus->pmus[topology_logical_die_id(cpu)] = pmu;
> +		rapl_pmus->pmus[rapl_pmu_idx] = pmu;
>  	}
>  
>  	/*
>  	 * Check if there is an online cpu in the package which collects rapl
>  	 * events already.
>  	 */
> -	target = cpumask_any_and(&rapl_cpu_mask, topology_die_cpumask(cpu));
> +	target = cpumask_any_and(&rapl_cpu_mask, rapl_pmu_cpumask);
>  	if (target < nr_cpu_ids)
>  		return 0;
>  
> @@ -675,7 +701,10 @@ static const struct attribute_group *rapl_attr_update[] = {
>  
>  static int __init init_rapl_pmus(void)
>  {
> -	int nr_rapl_pmu = topology_max_packages() * topology_max_dies_per_package();
> +	int nr_rapl_pmu = topology_max_packages();
> +
> +	if (!rapl_pmu_is_pkg_scope())
> +		nr_rapl_pmu *= topology_max_dies_per_package();
>  
>  	rapl_pmus = kzalloc(struct_size(rapl_pmus, pmus, nr_rapl_pmu), GFP_KERNEL);
>  	if (!rapl_pmus)
diff mbox series

Patch

diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
index 0c5e7a7c43ac..df71f38ad98d 100644
--- a/arch/x86/events/rapl.c
+++ b/arch/x86/events/rapl.c
@@ -103,6 +103,13 @@  static struct perf_pmu_events_attr event_attr_##v = {				\
 	.event_str	= str,							\
 };
 
+/*
+ * RAPL PMU scope for AMD is package whereas for Intel it is die.
+ */
+#define rapl_pmu_is_pkg_scope()				\
+	(boot_cpu_data.x86_vendor == X86_VENDOR_AMD ||	\
+	 boot_cpu_data.x86_vendor == X86_VENDOR_HYGON)
+
 struct rapl_pmu {
 	raw_spinlock_t		lock;
 	int			n_active;
@@ -140,9 +147,25 @@  static unsigned int rapl_cntr_mask;
 static u64 rapl_timer_ms;
 static struct perf_msr *rapl_msrs;
 
+/*
+ * Helper functions to get the correct topology macros according to the
+ * RAPL PMU scope.
+ */
+static inline unsigned int get_rapl_pmu_idx(int cpu)
+{
+	return rapl_pmu_is_pkg_scope() ? topology_logical_package_id(cpu) :
+					 topology_logical_die_id(cpu);
+}
+
+static inline const struct cpumask *get_rapl_pmu_cpumask(int cpu)
+{
+	return rapl_pmu_is_pkg_scope() ? topology_core_cpumask(cpu) :
+					 topology_die_cpumask(cpu);
+}
+
 static inline struct rapl_pmu *cpu_to_rapl_pmu(unsigned int cpu)
 {
-	unsigned int rapl_pmu_idx = topology_logical_die_id(cpu);
+	unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
 
 	/*
 	 * The unsigned check also catches the '-1' return value for non
@@ -543,6 +566,7 @@  static struct perf_msr amd_rapl_msrs[] = {
 
 static int rapl_cpu_offline(unsigned int cpu)
 {
+	const struct cpumask *rapl_pmu_cpumask = get_rapl_pmu_cpumask(cpu);
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
 
@@ -552,7 +576,7 @@  static int rapl_cpu_offline(unsigned int cpu)
 
 	pmu->cpu = -1;
 	/* Find a new cpu to collect rapl events */
-	target = cpumask_any_but(topology_die_cpumask(cpu), cpu);
+	target = cpumask_any_but(rapl_pmu_cpumask, cpu);
 
 	/* Migrate rapl events to the new target */
 	if (target < nr_cpu_ids) {
@@ -565,6 +589,8 @@  static int rapl_cpu_offline(unsigned int cpu)
 
 static int rapl_cpu_online(unsigned int cpu)
 {
+	unsigned int rapl_pmu_idx = get_rapl_pmu_idx(cpu);
+	const struct cpumask *rapl_pmu_cpumask = get_rapl_pmu_cpumask(cpu);
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
 
@@ -579,14 +605,14 @@  static int rapl_cpu_online(unsigned int cpu)
 		pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
 		rapl_hrtimer_init(pmu);
 
-		rapl_pmus->pmus[topology_logical_die_id(cpu)] = pmu;
+		rapl_pmus->pmus[rapl_pmu_idx] = pmu;
 	}
 
 	/*
 	 * Check if there is an online cpu in the package which collects rapl
 	 * events already.
 	 */
-	target = cpumask_any_and(&rapl_cpu_mask, topology_die_cpumask(cpu));
+	target = cpumask_any_and(&rapl_cpu_mask, rapl_pmu_cpumask);
 	if (target < nr_cpu_ids)
 		return 0;
 
@@ -675,7 +701,10 @@  static const struct attribute_group *rapl_attr_update[] = {
 
 static int __init init_rapl_pmus(void)
 {
-	int nr_rapl_pmu = topology_max_packages() * topology_max_dies_per_package();
+	int nr_rapl_pmu = topology_max_packages();
+
+	if (!rapl_pmu_is_pkg_scope())
+		nr_rapl_pmu *= topology_max_dies_per_package();
 
 	rapl_pmus = kzalloc(struct_size(rapl_pmus, pmus, nr_rapl_pmu), GFP_KERNEL);
 	if (!rapl_pmus)