[PATCH V5 0/2] cpufreq: cppc: Add support for frequency invariance

[PATCH V5 0/2] cpufreq: cppc: Add support for frequency invariance

[Viresh Kumar]

Hello,

CPPC cpufreq driver is used for ARM servers and this patch series tries
to provide counter-based frequency invariance support for them in the
absence for architecture specific counters (like AMUs).

This is tested by:
- /me with some hacks on Hikey, as I didn't have access to the right
  hardware.

- Vincent Guittot on ThunderX2, only initial testing done.

- Ionela Voinescu on Juno R2, though she tested a previous version of
  this.


This is based of 5.12-rc1.

Changes since V4:
- Move some code to policy specific initialization for cppc driver.
- Initialize kthread specific stuff only once in cppc driver.
- Added a kerneldoc comment in cppc driver and improved changelog as
  well.

Changes since V3:
- rebuild_sched_domains_energy() stuff moved from arm64 to drivers/base.
- Added Reviewed/Tested-by Ionela for the first patch.
- Remove unused max_freq field from structure in cppc driver.
- s/cppc_f_i/cppc_freq_inv.
- Fix an per-cpu access, there was a bug in earlier version.
- Create a single kthread which can run on any CPU and takes care of
  work from all the CPUs.
- Do the whole FIE thing under a new CONFIG option for cppc driver.
- Few minor improvements.

Changes since V2:
- Not sending as an RFC anymore.
- Several renames, reordering of code in 1/2 based on Ionela's comments.
- Several rebase changes for 2/2.
- The freq_scale calculations are optimized a bit.
- Better overall commenting and commit logs.

Changes since V1:
- The interface for setting the callbacks is improved, so different
  parts looking to provide their callbacks don't need to think about
  each other.

- Moved to per-cpu storage for storing the callback related data, AMU
  counters have higher priority with this.

--
viresh

Viresh Kumar (2):
  topology: Allow multiple entities to provide sched_freq_tick()
    callback
  cpufreq: CPPC: Add support for frequency invariance

 arch/arm64/include/asm/topology.h |  10 +-
 arch/arm64/kernel/topology.c      | 105 +++++--------
 drivers/base/arch_topology.c      |  85 ++++++++++-
 drivers/cpufreq/Kconfig.arm       |   9 ++
 drivers/cpufreq/cppc_cpufreq.c    | 244 ++++++++++++++++++++++++++++--
 include/linux/arch_topology.h     |  15 +-
 kernel/sched/core.c               |   1 +
 7 files changed, 377 insertions(+), 92 deletions(-)


base-commit: fe07bfda2fb9cdef8a4d4008a409bb02f35f1bd8
-- 
2.25.0.rc1.19.g042ed3e048af

[PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Viresh Kumar]

This patch attempts to make it generic enough so other parts of the
kernel can also provide their own implementation of scale_freq_tick()
callback, which is called by the scheduler periodically to update the
per-cpu freq_scale variable.

The implementations now need to provide 'struct scale_freq_data' for the
CPUs for which they have hardware counters available, and a callback
gets registered for each possible CPU in a per-cpu variable.

The arch specific (or ARM AMU) counters are updated to adapt to this and
they take the highest priority if they are available, i.e. they will be
used instead of CPPC based counters for example.

The special code to rebuild the sched domains, in case invariance status
change for the system, is moved out of arm64 specific code and is added
to arch_topology.c.

Note that this also defines SCALE_FREQ_SOURCE_CPUFREQ but doesn't use it
and it is added to show that cpufreq is also acts as source of
information for FIE and will be used by default if no other counters are
supported for a platform.

Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com>
Tested-by: Ionela Voinescu <ionela.voinescu@arm.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 arch/arm64/include/asm/topology.h |  10 +--
 arch/arm64/kernel/topology.c      | 105 +++++++++++-------------------
 drivers/base/arch_topology.c      |  85 ++++++++++++++++++++++--
 include/linux/arch_topology.h     |  14 +++-
 4 files changed, 134 insertions(+), 80 deletions(-)

diff --git a/arch/arm64/include/asm/topology.h b/arch/arm64/include/asm/topology.h
index 3b8dca4eb08d..ec2db3419c41 100644
--- a/arch/arm64/include/asm/topology.h
+++ b/arch/arm64/include/asm/topology.h
@@ -17,17 +17,9 @@ int pcibus_to_node(struct pci_bus *bus);
 #include <linux/arch_topology.h>
 
 void update_freq_counters_refs(void);
-void topology_scale_freq_tick(void);
-
-#ifdef CONFIG_ARM64_AMU_EXTN
-/*
- * Replace task scheduler's default counter-based
- * frequency-invariance scale factor setting.
- */
-#define arch_scale_freq_tick topology_scale_freq_tick
-#endif /* CONFIG_ARM64_AMU_EXTN */
 
 /* Replace task scheduler's default frequency-invariant accounting */
+#define arch_scale_freq_tick topology_scale_freq_tick
 #define arch_set_freq_scale topology_set_freq_scale
 #define arch_scale_freq_capacity topology_get_freq_scale
 #define arch_scale_freq_invariant topology_scale_freq_invariant
diff --git a/arch/arm64/kernel/topology.c b/arch/arm64/kernel/topology.c
index e08a4126453a..47fca7376c93 100644
--- a/arch/arm64/kernel/topology.c
+++ b/arch/arm64/kernel/topology.c
@@ -199,12 +199,47 @@ static int freq_inv_set_max_ratio(int cpu, u64 max_rate, u64 ref_rate)
 	return 0;
 }
 
-static DEFINE_STATIC_KEY_FALSE(amu_fie_key);
-#define amu_freq_invariant() static_branch_unlikely(&amu_fie_key)
+static void amu_scale_freq_tick(void)
+{
+	u64 prev_core_cnt, prev_const_cnt;
+	u64 core_cnt, const_cnt, scale;
+
+	prev_const_cnt = this_cpu_read(arch_const_cycles_prev);
+	prev_core_cnt = this_cpu_read(arch_core_cycles_prev);
+
+	update_freq_counters_refs();
+
+	const_cnt = this_cpu_read(arch_const_cycles_prev);
+	core_cnt = this_cpu_read(arch_core_cycles_prev);
+
+	if (unlikely(core_cnt <= prev_core_cnt ||
+		     const_cnt <= prev_const_cnt))
+		return;
+
+	/*
+	 *	    /\core    arch_max_freq_scale
+	 * scale =  ------- * --------------------
+	 *	    /\const   SCHED_CAPACITY_SCALE
+	 *
+	 * See validate_cpu_freq_invariance_counters() for details on
+	 * arch_max_freq_scale and the use of SCHED_CAPACITY_SHIFT.
+	 */
+	scale = core_cnt - prev_core_cnt;
+	scale *= this_cpu_read(arch_max_freq_scale);
+	scale = div64_u64(scale >> SCHED_CAPACITY_SHIFT,
+			  const_cnt - prev_const_cnt);
+
+	scale = min_t(unsigned long, scale, SCHED_CAPACITY_SCALE);
+	this_cpu_write(freq_scale, (unsigned long)scale);
+}
+
+static struct scale_freq_data amu_sfd = {
+	.source = SCALE_FREQ_SOURCE_ARCH,
+	.set_freq_scale = amu_scale_freq_tick,
+};
 
 static void amu_fie_setup(const struct cpumask *cpus)
 {
-	bool invariant;
 	int cpu;
 
 	/* We are already set since the last insmod of cpufreq driver */
@@ -221,25 +256,10 @@ static void amu_fie_setup(const struct cpumask *cpus)
 
 	cpumask_or(amu_fie_cpus, amu_fie_cpus, cpus);
 
-	invariant = topology_scale_freq_invariant();
-
-	/* We aren't fully invariant yet */
-	if (!invariant && !cpumask_equal(amu_fie_cpus, cpu_present_mask))
-		return;
-
-	static_branch_enable(&amu_fie_key);
+	topology_set_scale_freq_source(&amu_sfd, amu_fie_cpus);
 
 	pr_debug("CPUs[%*pbl]: counters will be used for FIE.",
 		 cpumask_pr_args(cpus));
-
-	/*
-	 * Task scheduler behavior depends on frequency invariance support,
-	 * either cpufreq or counter driven. If the support status changes as
-	 * a result of counter initialisation and use, retrigger the build of
-	 * scheduling domains to ensure the information is propagated properly.
-	 */
-	if (!invariant)
-		rebuild_sched_domains_energy();
 }
 
 static int init_amu_fie_callback(struct notifier_block *nb, unsigned long val,
@@ -283,53 +303,6 @@ static int __init init_amu_fie(void)
 }
 core_initcall(init_amu_fie);
 
-bool arch_freq_counters_available(const struct cpumask *cpus)
-{
-	return amu_freq_invariant() &&
-	       cpumask_subset(cpus, amu_fie_cpus);
-}
-
-void topology_scale_freq_tick(void)
-{
-	u64 prev_core_cnt, prev_const_cnt;
-	u64 core_cnt, const_cnt, scale;
-	int cpu = smp_processor_id();
-
-	if (!amu_freq_invariant())
-		return;
-
-	if (!cpumask_test_cpu(cpu, amu_fie_cpus))
-		return;
-
-	prev_const_cnt = this_cpu_read(arch_const_cycles_prev);
-	prev_core_cnt = this_cpu_read(arch_core_cycles_prev);
-
-	update_freq_counters_refs();
-
-	const_cnt = this_cpu_read(arch_const_cycles_prev);
-	core_cnt = this_cpu_read(arch_core_cycles_prev);
-
-	if (unlikely(core_cnt <= prev_core_cnt ||
-		     const_cnt <= prev_const_cnt))
-		return;
-
-	/*
-	 *	    /\core    arch_max_freq_scale
-	 * scale =  ------- * --------------------
-	 *	    /\const   SCHED_CAPACITY_SCALE
-	 *
-	 * See validate_cpu_freq_invariance_counters() for details on
-	 * arch_max_freq_scale and the use of SCHED_CAPACITY_SHIFT.
-	 */
-	scale = core_cnt - prev_core_cnt;
-	scale *= this_cpu_read(arch_max_freq_scale);
-	scale = div64_u64(scale >> SCHED_CAPACITY_SHIFT,
-			  const_cnt - prev_const_cnt);
-
-	scale = min_t(unsigned long, scale, SCHED_CAPACITY_SCALE);
-	this_cpu_write(freq_scale, (unsigned long)scale);
-}
-
 #ifdef CONFIG_ACPI_CPPC_LIB
 #include <acpi/cppc_acpi.h>
 
diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
index de8587cc119e..8f62dbf93f67 100644
--- a/drivers/base/arch_topology.c
+++ b/drivers/base/arch_topology.c
@@ -21,17 +21,94 @@
 #include <linux/sched.h>
 #include <linux/smp.h>
 
+static DEFINE_PER_CPU(struct scale_freq_data *, sft_data);
+static struct cpumask scale_freq_counters_mask;
+static bool scale_freq_invariant;
+
+static bool supports_scale_freq_counters(const struct cpumask *cpus)
+{
+	return cpumask_subset(cpus, &scale_freq_counters_mask);
+}
+
 bool topology_scale_freq_invariant(void)
 {
 	return cpufreq_supports_freq_invariance() ||
-	       arch_freq_counters_available(cpu_online_mask);
+	       supports_scale_freq_counters(cpu_online_mask);
 }
 
-__weak bool arch_freq_counters_available(const struct cpumask *cpus)
+static void update_scale_freq_invariant(bool status)
 {
-	return false;
+	if (scale_freq_invariant == status)
+		return;
+
+	/*
+	 * Task scheduler behavior depends on frequency invariance support,
+	 * either cpufreq or counter driven. If the support status changes as
+	 * a result of counter initialisation and use, retrigger the build of
+	 * scheduling domains to ensure the information is propagated properly.
+	 */
+	if (topology_scale_freq_invariant() == status) {
+		scale_freq_invariant = status;
+		rebuild_sched_domains_energy();
+	}
 }
+
+void topology_set_scale_freq_source(struct scale_freq_data *data,
+				    const struct cpumask *cpus)
+{
+	struct scale_freq_data *sfd;
+	int cpu;
+
+	/*
+	 * Avoid calling rebuild_sched_domains() unnecessarily if FIE is
+	 * supported by cpufreq.
+	 */
+	if (cpumask_empty(&scale_freq_counters_mask))
+		scale_freq_invariant = topology_scale_freq_invariant();
+
+	for_each_cpu(cpu, cpus) {
+		sfd = per_cpu(sft_data, cpu);
+
+		/* Use ARCH provided counters whenever possible */
+		if (!sfd || sfd->source != SCALE_FREQ_SOURCE_ARCH) {
+			per_cpu(sft_data, cpu) = data;
+			cpumask_set_cpu(cpu, &scale_freq_counters_mask);
+		}
+	}
+
+	update_scale_freq_invariant(true);
+}
+EXPORT_SYMBOL_GPL(topology_set_scale_freq_source);
+
+void topology_clear_scale_freq_source(enum scale_freq_source source,
+				      const struct cpumask *cpus)
+{
+	struct scale_freq_data *sfd;
+	int cpu;
+
+	for_each_cpu(cpu, cpus) {
+		sfd = per_cpu(sft_data, cpu);
+
+		if (sfd && sfd->source == source) {
+			per_cpu(sft_data, cpu) = NULL;
+			cpumask_clear_cpu(cpu, &scale_freq_counters_mask);
+		}
+	}
+
+	update_scale_freq_invariant(false);
+}
+EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);
+
+void topology_scale_freq_tick(void)
+{
+	struct scale_freq_data *sfd = *this_cpu_ptr(&sft_data);
+
+	if (sfd)
+		sfd->set_freq_scale();
+}
+
 DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;
+EXPORT_SYMBOL_GPL(freq_scale);
 
 void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
 			     unsigned long max_freq)
@@ -47,7 +124,7 @@ void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
 	 * want to update the scale factor with information from CPUFREQ.
 	 * Instead the scale factor will be updated from arch_scale_freq_tick.
 	 */
-	if (arch_freq_counters_available(cpus))
+	if (supports_scale_freq_counters(cpus))
 		return;
 
 	scale = (cur_freq << SCHED_CAPACITY_SHIFT) / max_freq;
diff --git a/include/linux/arch_topology.h b/include/linux/arch_topology.h
index 0f6cd6b73a61..3bcfba5c21a7 100644
--- a/include/linux/arch_topology.h
+++ b/include/linux/arch_topology.h
@@ -34,7 +34,19 @@ void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
 			     unsigned long max_freq);
 bool topology_scale_freq_invariant(void);
 
-bool arch_freq_counters_available(const struct cpumask *cpus);
+enum scale_freq_source {
+	SCALE_FREQ_SOURCE_CPUFREQ = 0,
+	SCALE_FREQ_SOURCE_ARCH,
+};
+
+struct scale_freq_data {
+	enum scale_freq_source source;
+	void (*set_freq_scale)(void);
+};
+
+void topology_scale_freq_tick(void);
+void topology_set_scale_freq_source(struct scale_freq_data *data, const struct cpumask *cpus);
+void topology_clear_scale_freq_source(enum scale_freq_source source, const struct cpumask *cpus);
 
 DECLARE_PER_CPU(unsigned long, thermal_pressure);
 
-- 
2.25.0.rc1.19.g042ed3e048af

Re: [PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Will Deacon]

On Mon, Mar 01, 2021 at 12:21:17PM +0530, Viresh Kumar wrote:
> This patch attempts to make it generic enough so other parts of the
> kernel can also provide their own implementation of scale_freq_tick()
> callback, which is called by the scheduler periodically to update the
> per-cpu freq_scale variable.
> 
> The implementations now need to provide 'struct scale_freq_data' for the
> CPUs for which they have hardware counters available, and a callback
> gets registered for each possible CPU in a per-cpu variable.
> 
> The arch specific (or ARM AMU) counters are updated to adapt to this and
> they take the highest priority if they are available, i.e. they will be
> used instead of CPPC based counters for example.
> 
> The special code to rebuild the sched domains, in case invariance status
> change for the system, is moved out of arm64 specific code and is added
> to arch_topology.c.
> 
> Note that this also defines SCALE_FREQ_SOURCE_CPUFREQ but doesn't use it
> and it is added to show that cpufreq is also acts as source of
> information for FIE and will be used by default if no other counters are
> supported for a platform.
> 
> Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Tested-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> ---
>  arch/arm64/include/asm/topology.h |  10 +--
>  arch/arm64/kernel/topology.c      | 105 +++++++++++-------------------
>  drivers/base/arch_topology.c      |  85 ++++++++++++++++++++++--
>  include/linux/arch_topology.h     |  14 +++-
>  4 files changed, 134 insertions(+), 80 deletions(-)

For the arm64 bits:

Acked-by: Will Deacon <will@kernel.org>

However...

> diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
> index de8587cc119e..8f62dbf93f67 100644
> --- a/drivers/base/arch_topology.c
> +++ b/drivers/base/arch_topology.c
> @@ -21,17 +21,94 @@
>  #include <linux/sched.h>
>  #include <linux/smp.h>
>  
> +static DEFINE_PER_CPU(struct scale_freq_data *, sft_data);
> +static struct cpumask scale_freq_counters_mask;
> +static bool scale_freq_invariant;
> +
> +static bool supports_scale_freq_counters(const struct cpumask *cpus)
> +{
> +	return cpumask_subset(cpus, &scale_freq_counters_mask);
> +}
> +
>  bool topology_scale_freq_invariant(void)
>  {
>  	return cpufreq_supports_freq_invariance() ||
> -	       arch_freq_counters_available(cpu_online_mask);
> +	       supports_scale_freq_counters(cpu_online_mask);
>  }
>  
> -__weak bool arch_freq_counters_available(const struct cpumask *cpus)
> +static void update_scale_freq_invariant(bool status)
>  {
> -	return false;
> +	if (scale_freq_invariant == status)
> +		return;
> +
> +	/*
> +	 * Task scheduler behavior depends on frequency invariance support,
> +	 * either cpufreq or counter driven. If the support status changes as
> +	 * a result of counter initialisation and use, retrigger the build of
> +	 * scheduling domains to ensure the information is propagated properly.
> +	 */
> +	if (topology_scale_freq_invariant() == status) {
> +		scale_freq_invariant = status;
> +		rebuild_sched_domains_energy();
> +	}
>  }
> +
> +void topology_set_scale_freq_source(struct scale_freq_data *data,
> +				    const struct cpumask *cpus)
> +{
> +	struct scale_freq_data *sfd;
> +	int cpu;
> +
> +	/*
> +	 * Avoid calling rebuild_sched_domains() unnecessarily if FIE is
> +	 * supported by cpufreq.
> +	 */
> +	if (cpumask_empty(&scale_freq_counters_mask))
> +		scale_freq_invariant = topology_scale_freq_invariant();
> +
> +	for_each_cpu(cpu, cpus) {
> +		sfd = per_cpu(sft_data, cpu);
> +
> +		/* Use ARCH provided counters whenever possible */
> +		if (!sfd || sfd->source != SCALE_FREQ_SOURCE_ARCH) {
> +			per_cpu(sft_data, cpu) = data;
> +			cpumask_set_cpu(cpu, &scale_freq_counters_mask);
> +		}
> +	}
> +
> +	update_scale_freq_invariant(true);
> +}
> +EXPORT_SYMBOL_GPL(topology_set_scale_freq_source);

I don't get why you need to export this in this patch. The arm64 topology
code is never built as a module.

> +
> +void topology_clear_scale_freq_source(enum scale_freq_source source,
> +				      const struct cpumask *cpus)
> +{
> +	struct scale_freq_data *sfd;
> +	int cpu;
> +
> +	for_each_cpu(cpu, cpus) {
> +		sfd = per_cpu(sft_data, cpu);
> +
> +		if (sfd && sfd->source == source) {
> +			per_cpu(sft_data, cpu) = NULL;
> +			cpumask_clear_cpu(cpu, &scale_freq_counters_mask);
> +		}
> +	}
> +
> +	update_scale_freq_invariant(false);
> +}
> +EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);

Same here.

> +
> +void topology_scale_freq_tick(void)
> +{
> +	struct scale_freq_data *sfd = *this_cpu_ptr(&sft_data);
> +
> +	if (sfd)
> +		sfd->set_freq_scale();
> +}
> +
>  DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;
> +EXPORT_SYMBOL_GPL(freq_scale);

And here. This one probably wants a less generic name as well if it's going
to be exported.

Will

Re: [PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Viresh Kumar]

On 08-03-21, 14:52, Will Deacon wrote:
> On Mon, Mar 01, 2021 at 12:21:17PM +0530, Viresh Kumar wrote:
> > +EXPORT_SYMBOL_GPL(topology_set_scale_freq_source);
> 
> I don't get why you need to export this in this patch. The arm64 topology
> code is never built as a module.
> 
> > +EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);
> 
> Same here.
> 
> > +EXPORT_SYMBOL_GPL(freq_scale);
> 
> And here.

After this patch, any part of the kernel can use these
helpers/variables to run their own implementation of tick-freq-scale
and so this patch looked to be the right place for that to me.

And the second patch in the series updates the CPPC cpufreq driver
(tristate) to use these exported symbols, so we have the first user
who needs the exported symbols as well.

> This one probably wants a less generic name as well if it's going
> to be exported.

x86 names it arch_freq_scale, perhaps we should stick to that instead.

-- 
viresh

Re: [PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Will Deacon]

On Tue, Mar 09, 2021 at 09:46:43AM +0530, Viresh Kumar wrote:
> On 08-03-21, 14:52, Will Deacon wrote:
> > On Mon, Mar 01, 2021 at 12:21:17PM +0530, Viresh Kumar wrote:
> > > +EXPORT_SYMBOL_GPL(topology_set_scale_freq_source);
> > 
> > I don't get why you need to export this in this patch. The arm64 topology
> > code is never built as a module.
> > 
> > > +EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);
> > 
> > Same here.
> > 
> > > +EXPORT_SYMBOL_GPL(freq_scale);
> > 
> > And here.
> 
> After this patch, any part of the kernel can use these
> helpers/variables to run their own implementation of tick-freq-scale
> and so this patch looked to be the right place for that to me.
> 
> And the second patch in the series updates the CPPC cpufreq driver
> (tristate) to use these exported symbols, so we have the first user
> who needs the exported symbols as well.

Ok, then how about putting the exports in their own patch that immediately
precedes the patch which uses them? I think that would make it much clearer.

> > This one probably wants a less generic name as well if it's going
> > to be exported.
> 
> x86 names it arch_freq_scale, perhaps we should stick to that instead.

Sounds like a improvement, thanks.

Will

Re: [PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Ionela Voinescu]

On Monday 01 Mar 2021 at 12:21:17 (+0530), Viresh Kumar wrote:
> This patch attempts to make it generic enough so other parts of the
> kernel can also provide their own implementation of scale_freq_tick()
> callback, which is called by the scheduler periodically to update the
> per-cpu freq_scale variable.
> 
> The implementations now need to provide 'struct scale_freq_data' for the
> CPUs for which they have hardware counters available, and a callback
> gets registered for each possible CPU in a per-cpu variable.
> 
> The arch specific (or ARM AMU) counters are updated to adapt to this and
> they take the highest priority if they are available, i.e. they will be
> used instead of CPPC based counters for example.
> 
> The special code to rebuild the sched domains, in case invariance status
> change for the system, is moved out of arm64 specific code and is added
> to arch_topology.c.
> 
> Note that this also defines SCALE_FREQ_SOURCE_CPUFREQ but doesn't use it
> and it is added to show that cpufreq is also acts as source of
> information for FIE and will be used by default if no other counters are
> supported for a platform.
> 
> Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Tested-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> ---
>  arch/arm64/include/asm/topology.h |  10 +--
>  arch/arm64/kernel/topology.c      | 105 +++++++++++-------------------
>  drivers/base/arch_topology.c      |  85 ++++++++++++++++++++++--
>  include/linux/arch_topology.h     |  14 +++-
>  4 files changed, 134 insertions(+), 80 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/topology.h b/arch/arm64/include/asm/topology.h
> index 3b8dca4eb08d..ec2db3419c41 100644
> --- a/arch/arm64/include/asm/topology.h
> +++ b/arch/arm64/include/asm/topology.h
> @@ -17,17 +17,9 @@ int pcibus_to_node(struct pci_bus *bus);
>  #include <linux/arch_topology.h>
>  
>  void update_freq_counters_refs(void);
> -void topology_scale_freq_tick(void);
> -
> -#ifdef CONFIG_ARM64_AMU_EXTN
> -/*
> - * Replace task scheduler's default counter-based
> - * frequency-invariance scale factor setting.
> - */
> -#define arch_scale_freq_tick topology_scale_freq_tick
> -#endif /* CONFIG_ARM64_AMU_EXTN */
>  
>  /* Replace task scheduler's default frequency-invariant accounting */
> +#define arch_scale_freq_tick topology_scale_freq_tick
>  #define arch_set_freq_scale topology_set_freq_scale
>  #define arch_scale_freq_capacity topology_get_freq_scale
>  #define arch_scale_freq_invariant topology_scale_freq_invariant
> diff --git a/arch/arm64/kernel/topology.c b/arch/arm64/kernel/topology.c
> index e08a4126453a..47fca7376c93 100644
> --- a/arch/arm64/kernel/topology.c
> +++ b/arch/arm64/kernel/topology.c
> @@ -199,12 +199,47 @@ static int freq_inv_set_max_ratio(int cpu, u64 max_rate, u64 ref_rate)
>  	return 0;
>  }
>  
> -static DEFINE_STATIC_KEY_FALSE(amu_fie_key);
> -#define amu_freq_invariant() static_branch_unlikely(&amu_fie_key)
> +static void amu_scale_freq_tick(void)
> +{
> +	u64 prev_core_cnt, prev_const_cnt;
> +	u64 core_cnt, const_cnt, scale;
> +
> +	prev_const_cnt = this_cpu_read(arch_const_cycles_prev);
> +	prev_core_cnt = this_cpu_read(arch_core_cycles_prev);
> +
> +	update_freq_counters_refs();
> +
> +	const_cnt = this_cpu_read(arch_const_cycles_prev);
> +	core_cnt = this_cpu_read(arch_core_cycles_prev);
> +
> +	if (unlikely(core_cnt <= prev_core_cnt ||
> +		     const_cnt <= prev_const_cnt))
> +		return;
> +
> +	/*
> +	 *	    /\core    arch_max_freq_scale
> +	 * scale =  ------- * --------------------
> +	 *	    /\const   SCHED_CAPACITY_SCALE
> +	 *
> +	 * See validate_cpu_freq_invariance_counters() for details on
> +	 * arch_max_freq_scale and the use of SCHED_CAPACITY_SHIFT.
> +	 */
> +	scale = core_cnt - prev_core_cnt;
> +	scale *= this_cpu_read(arch_max_freq_scale);
> +	scale = div64_u64(scale >> SCHED_CAPACITY_SHIFT,
> +			  const_cnt - prev_const_cnt);
> +
> +	scale = min_t(unsigned long, scale, SCHED_CAPACITY_SCALE);
> +	this_cpu_write(freq_scale, (unsigned long)scale);
> +}
> +
> +static struct scale_freq_data amu_sfd = {
> +	.source = SCALE_FREQ_SOURCE_ARCH,
> +	.set_freq_scale = amu_scale_freq_tick,
> +};
>  
>  static void amu_fie_setup(const struct cpumask *cpus)
>  {
> -	bool invariant;
>  	int cpu;
>  
>  	/* We are already set since the last insmod of cpufreq driver */
> @@ -221,25 +256,10 @@ static void amu_fie_setup(const struct cpumask *cpus)
>  
>  	cpumask_or(amu_fie_cpus, amu_fie_cpus, cpus);
>  
> -	invariant = topology_scale_freq_invariant();
> -
> -	/* We aren't fully invariant yet */
> -	if (!invariant && !cpumask_equal(amu_fie_cpus, cpu_present_mask))
> -		return;
> -
> -	static_branch_enable(&amu_fie_key);
> +	topology_set_scale_freq_source(&amu_sfd, amu_fie_cpus);
>  
>  	pr_debug("CPUs[%*pbl]: counters will be used for FIE.",
>  		 cpumask_pr_args(cpus));
> -
> -	/*
> -	 * Task scheduler behavior depends on frequency invariance support,
> -	 * either cpufreq or counter driven. If the support status changes as
> -	 * a result of counter initialisation and use, retrigger the build of
> -	 * scheduling domains to ensure the information is propagated properly.
> -	 */
> -	if (!invariant)
> -		rebuild_sched_domains_energy();
>  }
>  
>  static int init_amu_fie_callback(struct notifier_block *nb, unsigned long val,
> @@ -283,53 +303,6 @@ static int __init init_amu_fie(void)
>  }
>  core_initcall(init_amu_fie);
>  
> -bool arch_freq_counters_available(const struct cpumask *cpus)
> -{
> -	return amu_freq_invariant() &&
> -	       cpumask_subset(cpus, amu_fie_cpus);
> -}
> -
> -void topology_scale_freq_tick(void)
> -{
> -	u64 prev_core_cnt, prev_const_cnt;
> -	u64 core_cnt, const_cnt, scale;
> -	int cpu = smp_processor_id();
> -
> -	if (!amu_freq_invariant())
> -		return;
> -
> -	if (!cpumask_test_cpu(cpu, amu_fie_cpus))
> -		return;
> -
> -	prev_const_cnt = this_cpu_read(arch_const_cycles_prev);
> -	prev_core_cnt = this_cpu_read(arch_core_cycles_prev);
> -
> -	update_freq_counters_refs();
> -
> -	const_cnt = this_cpu_read(arch_const_cycles_prev);
> -	core_cnt = this_cpu_read(arch_core_cycles_prev);
> -
> -	if (unlikely(core_cnt <= prev_core_cnt ||
> -		     const_cnt <= prev_const_cnt))
> -		return;
> -
> -	/*
> -	 *	    /\core    arch_max_freq_scale
> -	 * scale =  ------- * --------------------
> -	 *	    /\const   SCHED_CAPACITY_SCALE
> -	 *
> -	 * See validate_cpu_freq_invariance_counters() for details on
> -	 * arch_max_freq_scale and the use of SCHED_CAPACITY_SHIFT.
> -	 */
> -	scale = core_cnt - prev_core_cnt;
> -	scale *= this_cpu_read(arch_max_freq_scale);
> -	scale = div64_u64(scale >> SCHED_CAPACITY_SHIFT,
> -			  const_cnt - prev_const_cnt);
> -
> -	scale = min_t(unsigned long, scale, SCHED_CAPACITY_SCALE);
> -	this_cpu_write(freq_scale, (unsigned long)scale);
> -}
> -
>  #ifdef CONFIG_ACPI_CPPC_LIB
>  #include <acpi/cppc_acpi.h>
>  
> diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
> index de8587cc119e..8f62dbf93f67 100644
> --- a/drivers/base/arch_topology.c
> +++ b/drivers/base/arch_topology.c
> @@ -21,17 +21,94 @@
>  #include <linux/sched.h>
>  #include <linux/smp.h>
>  
> +static DEFINE_PER_CPU(struct scale_freq_data *, sft_data);
> +static struct cpumask scale_freq_counters_mask;
> +static bool scale_freq_invariant;
> +
> +static bool supports_scale_freq_counters(const struct cpumask *cpus)
> +{
> +	return cpumask_subset(cpus, &scale_freq_counters_mask);
> +}
> +
>  bool topology_scale_freq_invariant(void)
>  {
>  	return cpufreq_supports_freq_invariance() ||
> -	       arch_freq_counters_available(cpu_online_mask);
> +	       supports_scale_freq_counters(cpu_online_mask);
>  }
>  
> -__weak bool arch_freq_counters_available(const struct cpumask *cpus)
> +static void update_scale_freq_invariant(bool status)
>  {
> -	return false;
> +	if (scale_freq_invariant == status)
> +		return;
> +
> +	/*
> +	 * Task scheduler behavior depends on frequency invariance support,
> +	 * either cpufreq or counter driven. If the support status changes as
> +	 * a result of counter initialisation and use, retrigger the build of
> +	 * scheduling domains to ensure the information is propagated properly.
> +	 */
> +	if (topology_scale_freq_invariant() == status) {
> +		scale_freq_invariant = status;
> +		rebuild_sched_domains_energy();
> +	}
>  }
> +
> +void topology_set_scale_freq_source(struct scale_freq_data *data,
> +				    const struct cpumask *cpus)
> +{
> +	struct scale_freq_data *sfd;
> +	int cpu;
> +
> +	/*
> +	 * Avoid calling rebuild_sched_domains() unnecessarily if FIE is
> +	 * supported by cpufreq.
> +	 */
> +	if (cpumask_empty(&scale_freq_counters_mask))
> +		scale_freq_invariant = topology_scale_freq_invariant();
> +
> +	for_each_cpu(cpu, cpus) {
> +		sfd = per_cpu(sft_data, cpu);
> +
> +		/* Use ARCH provided counters whenever possible */
> +		if (!sfd || sfd->source != SCALE_FREQ_SOURCE_ARCH) {
> +			per_cpu(sft_data, cpu) = data;
> +			cpumask_set_cpu(cpu, &scale_freq_counters_mask);
> +		}
> +	}
> +
> +	update_scale_freq_invariant(true);
> +}
> +EXPORT_SYMBOL_GPL(topology_set_scale_freq_source);
> +
> +void topology_clear_scale_freq_source(enum scale_freq_source source,
> +				      const struct cpumask *cpus)
> +{
> +	struct scale_freq_data *sfd;
> +	int cpu;
> +
> +	for_each_cpu(cpu, cpus) {
> +		sfd = per_cpu(sft_data, cpu);
> +
> +		if (sfd && sfd->source == source) {
> +			per_cpu(sft_data, cpu) = NULL;
> +			cpumask_clear_cpu(cpu, &scale_freq_counters_mask);
> +		}
> +	}
> +
> +	update_scale_freq_invariant(false);
> +}
> +EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);
> +
> +void topology_scale_freq_tick(void)
> +{
> +	struct scale_freq_data *sfd = *this_cpu_ptr(&sft_data);
> +
> +	if (sfd)
> +		sfd->set_freq_scale();
> +}
> +
>  DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;
> +EXPORT_SYMBOL_GPL(freq_scale);
>  
>  void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
>  			     unsigned long max_freq)
> @@ -47,7 +124,7 @@ void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
>  	 * want to update the scale factor with information from CPUFREQ.
>  	 * Instead the scale factor will be updated from arch_scale_freq_tick.
>  	 */
> -	if (arch_freq_counters_available(cpus))
> +	if (supports_scale_freq_counters(cpus))
>  		return;
>  
>  	scale = (cur_freq << SCHED_CAPACITY_SHIFT) / max_freq;
> diff --git a/include/linux/arch_topology.h b/include/linux/arch_topology.h
> index 0f6cd6b73a61..3bcfba5c21a7 100644
> --- a/include/linux/arch_topology.h
> +++ b/include/linux/arch_topology.h
> @@ -34,7 +34,19 @@ void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
>  			     unsigned long max_freq);
>  bool topology_scale_freq_invariant(void);
>  
> -bool arch_freq_counters_available(const struct cpumask *cpus);
> +enum scale_freq_source {
> +	SCALE_FREQ_SOURCE_CPUFREQ = 0,
> +	SCALE_FREQ_SOURCE_ARCH,
> +};
> +
> +struct scale_freq_data {
> +	enum scale_freq_source source;
> +	void (*set_freq_scale)(void);
> +};
> +
> +void topology_scale_freq_tick(void);
> +void topology_set_scale_freq_source(struct scale_freq_data *data, const struct cpumask *cpus);
> +void topology_clear_scale_freq_source(enum scale_freq_source source, const struct cpumask *cpus);

Nit: can you split these lines?

Ionela.

>  
>  DECLARE_PER_CPU(unsigned long, thermal_pressure);
>  
> -- 
> 2.25.0.rc1.19.g042ed3e048af
> 

Re: [PATCH V5 1/2] topology: Allow multiple entities to provide sched_freq_tick() callback

[Viresh Kumar]

On 09-03-21, 15:11, Ionela Voinescu wrote:
> On Monday 01 Mar 2021 at 12:21:17 (+0530), Viresh Kumar wrote:
> > diff --git a/include/linux/arch_topology.h b/include/linux/arch_topology.h
> > index 0f6cd6b73a61..3bcfba5c21a7 100644
> > --- a/include/linux/arch_topology.h
> > +++ b/include/linux/arch_topology.h
> > @@ -34,7 +34,19 @@ void topology_set_freq_scale(const struct cpumask *cpus, unsigned long cur_freq,
> >  			     unsigned long max_freq);
> >  bool topology_scale_freq_invariant(void);

We don't normally add blank lines between declarations (variables or
function), like what's done here.

> >  
> > -bool arch_freq_counters_available(const struct cpumask *cpus);
> > +enum scale_freq_source {
> > +	SCALE_FREQ_SOURCE_CPUFREQ = 0,
> > +	SCALE_FREQ_SOURCE_ARCH,
> > +};
> > +
> > +struct scale_freq_data {
> > +	enum scale_freq_source source;
> > +	void (*set_freq_scale)(void);
> > +};
> > +
> > +void topology_scale_freq_tick(void);
> > +void topology_set_scale_freq_source(struct scale_freq_data *data, const struct cpumask *cpus);
> > +void topology_clear_scale_freq_source(enum scale_freq_source source, const struct cpumask *cpus);
> 
> Nit: can you split these lines?

And so in order to be consistent across the file, I haven't added one
here.

-- 
viresh

[PATCH V5 2/2] cpufreq: CPPC: Add support for frequency invariance

[Viresh Kumar]

The Frequency Invariance Engine (FIE) is providing a frequency scaling
correction factor that helps achieve more accurate load-tracking.

Normally, this scaling factor can be obtained directly with the help of
the cpufreq drivers as they know the exact frequency the hardware is
running at. But that isn't the case for CPPC cpufreq driver.

Another way of obtaining that is using the arch specific counter
support, which is already present in kernel, but that hardware is
optional for platforms.

This patch updates the CPPC driver to register itself with the topology
core to provide its own implementation (cppc_scale_freq_tick()) of
topology_scale_freq_tick() which gets called by the scheduler on every
tick. Note that the arch specific counters have higher priority than
CPPC counters, if available, though the CPPC driver doesn't need to have
any special handling for that.

On an invocation of cppc_scale_freq_tick(), we schedule an irq work
(since we reach here from hard-irq context), which then schedules a
normal work item and cppc_scale_freq_workfn() updates the per_cpu
freq_scale variable based on the counter updates since the last tick.

To allow platforms to disable frequency invariance support if they want,
this is all done under CONFIG_ACPI_CPPC_CPUFREQ_FIE, which is enabled by
default.

This also exports sched_setattr_nocheck() as the CPPC driver can be
built as a module.

Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: linux-acpi@vger.kernel.org
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/Kconfig.arm    |   9 ++
 drivers/cpufreq/cppc_cpufreq.c | 244 +++++++++++++++++++++++++++++++--
 include/linux/arch_topology.h  |   1 +
 kernel/sched/core.c            |   1 +
 4 files changed, 243 insertions(+), 12 deletions(-)

diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
index e65e0a43be64..a3e2d6dfea70 100644
--- a/drivers/cpufreq/Kconfig.arm
+++ b/drivers/cpufreq/Kconfig.arm
@@ -19,6 +19,15 @@ config ACPI_CPPC_CPUFREQ
 
 	  If in doubt, say N.
 
+config ACPI_CPPC_CPUFREQ_FIE
+	bool "Frequency Invariance support for CPPC cpufreq driver"
+	depends on ACPI_CPPC_CPUFREQ
+	default y
+	help
+	  This enables frequency invariance support for CPPC cpufreq driver.
+
+	  If in doubt, say N.
+
 config ARM_ALLWINNER_SUN50I_CPUFREQ_NVMEM
 	tristate "Allwinner nvmem based SUN50I CPUFreq driver"
 	depends on ARCH_SUNXI
diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
index 8a482c434ea6..c4580a37a1b1 100644
--- a/drivers/cpufreq/cppc_cpufreq.c
+++ b/drivers/cpufreq/cppc_cpufreq.c
@@ -10,14 +10,18 @@
 
 #define pr_fmt(fmt)	"CPPC Cpufreq:"	fmt
 
+#include <linux/arch_topology.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
 #include <linux/delay.h>
 #include <linux/cpu.h>
 #include <linux/cpufreq.h>
 #include <linux/dmi.h>
+#include <linux/irq_work.h>
+#include <linux/kthread.h>
 #include <linux/time.h>
 #include <linux/vmalloc.h>
+#include <uapi/linux/sched/types.h>
 
 #include <asm/unaligned.h>
 
@@ -57,6 +61,203 @@ static struct cppc_workaround_oem_info wa_info[] = {
 	}
 };
 
+#ifdef CONFIG_ACPI_CPPC_CPUFREQ_FIE
+
+/* Frequency invariance support */
+struct cppc_freq_invariance {
+	int cpu;
+	struct irq_work irq_work;
+	struct kthread_work work;
+	struct cppc_perf_fb_ctrs prev_perf_fb_ctrs;
+	struct cppc_cpudata *cpu_data;
+};
+
+static DEFINE_PER_CPU(struct cppc_freq_invariance, cppc_freq_inv);
+static struct kthread_worker *kworker_fie;
+
+static struct cpufreq_driver cppc_cpufreq_driver;
+static unsigned int hisi_cppc_cpufreq_get_rate(unsigned int cpu);
+static int cppc_perf_from_fbctrs(struct cppc_cpudata *cpu_data,
+				 struct cppc_perf_fb_ctrs fb_ctrs_t0,
+				 struct cppc_perf_fb_ctrs fb_ctrs_t1);
+
+/**
+ * cppc_scale_freq_workfn - CPPC freq_scale updater for frequency invariance
+ * @work: The work item.
+ *
+ * The CPPC driver register itself with the topology core to provide its own
+ * implementation (cppc_scale_freq_tick()) of topology_scale_freq_tick() which
+ * gets called by the scheduler on every tick.
+ *
+ * Note that the arch specific counters have higher priority than CPPC counters,
+ * if available, though the CPPC driver doesn't need to have any special
+ * handling for that.
+ *
+ * On an invocation of cppc_scale_freq_tick(), we schedule an irq work (since we
+ * reach here from hard-irq context), which then schedules a normal work item
+ * and cppc_scale_freq_workfn() updates the per_cpu freq_scale variable based on
+ * the counter updates since the last tick.
+ */
+static void cppc_scale_freq_workfn(struct kthread_work *work)
+{
+	struct cppc_freq_invariance *cppc_fi;
+	struct cppc_perf_fb_ctrs fb_ctrs = {0};
+	struct cppc_cpudata *cpu_data;
+	unsigned long local_freq_scale;
+	u64 perf;
+
+	cppc_fi = container_of(work, struct cppc_freq_invariance, work);
+	cpu_data = cppc_fi->cpu_data;
+
+	if (cppc_get_perf_ctrs(cppc_fi->cpu, &fb_ctrs)) {
+		pr_warn("%s: failed to read perf counters\n", __func__);
+		return;
+	}
+
+	cppc_fi->prev_perf_fb_ctrs = fb_ctrs;
+	perf = cppc_perf_from_fbctrs(cpu_data, cppc_fi->prev_perf_fb_ctrs,
+				     fb_ctrs);
+
+	perf <<= SCHED_CAPACITY_SHIFT;
+	local_freq_scale = div64_u64(perf, cpu_data->perf_caps.highest_perf);
+	if (WARN_ON(local_freq_scale > 1024))
+		local_freq_scale = 1024;
+
+	per_cpu(freq_scale, cppc_fi->cpu) = local_freq_scale;
+}
+
+static void cppc_irq_work(struct irq_work *irq_work)
+{
+	struct cppc_freq_invariance *cppc_fi;
+
+	cppc_fi = container_of(irq_work, struct cppc_freq_invariance, irq_work);
+	kthread_queue_work(kworker_fie, &cppc_fi->work);
+}
+
+static void cppc_scale_freq_tick(void)
+{
+	struct cppc_freq_invariance *cppc_fi = &per_cpu(cppc_freq_inv, smp_processor_id());
+
+	/*
+	 * cppc_get_perf_ctrs() can potentially sleep, call that from the right
+	 * context.
+	 */
+	irq_work_queue(&cppc_fi->irq_work);
+}
+
+static struct scale_freq_data cppc_sftd = {
+	.source = SCALE_FREQ_SOURCE_CPPC,
+	.set_freq_scale = cppc_scale_freq_tick,
+};
+
+static void cppc_freq_invariance_policy_init(struct cpufreq_policy *policy,
+					     struct cppc_cpudata *cpu_data)
+{
+	struct cppc_freq_invariance *cppc_fi;
+	int i;
+
+	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
+		return;
+
+	for_each_cpu(i, policy->cpus) {
+		cppc_fi = &per_cpu(cppc_freq_inv, i);
+		cppc_fi->cpu = i;
+		cppc_fi->cpu_data = cpu_data;
+		kthread_init_work(&cppc_fi->work, cppc_scale_freq_workfn);
+		init_irq_work(&cppc_fi->irq_work, cppc_irq_work);
+	}
+}
+
+static void cppc_freq_invariance_exit(void)
+{
+	struct cppc_freq_invariance *cppc_fi;
+	int i;
+
+	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
+		return;
+
+	topology_clear_scale_freq_source(SCALE_FREQ_SOURCE_CPPC, cpu_present_mask);
+
+	for_each_possible_cpu(i) {
+		cppc_fi = &per_cpu(cppc_freq_inv, i);
+		irq_work_sync(&cppc_fi->irq_work);
+	}
+
+	kthread_destroy_worker(kworker_fie);
+	kworker_fie = NULL;
+}
+
+static void __init cppc_freq_invariance_init(void)
+{
+	struct cppc_perf_fb_ctrs fb_ctrs = {0};
+	struct cppc_freq_invariance *cppc_fi;
+	struct sched_attr attr = {
+		.size		= sizeof(struct sched_attr),
+		.sched_policy	= SCHED_DEADLINE,
+		.sched_nice	= 0,
+		.sched_priority	= 0,
+		/*
+		 * Fake (unused) bandwidth; workaround to "fix"
+		 * priority inheritance.
+		 */
+		.sched_runtime	= 1000000,
+		.sched_deadline = 10000000,
+		.sched_period	= 10000000,
+	};
+	int i, ret;
+
+	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
+		return;
+
+	kworker_fie = kthread_create_worker(0, "cppc_fie");
+	if (IS_ERR(kworker_fie))
+		return;
+
+	ret = sched_setattr_nocheck(kworker_fie->task, &attr);
+	if (ret) {
+		pr_warn("%s: failed to set SCHED_DEADLINE: %d\n", __func__,
+			ret);
+		kthread_destroy_worker(kworker_fie);
+		return;
+	}
+
+	for_each_possible_cpu(i) {
+		cppc_fi = &per_cpu(cppc_freq_inv, i);
+
+		/* A policy failed to initialize, abort */
+		if (unlikely(!cppc_fi->cpu_data))
+			return cppc_freq_invariance_exit();
+
+		ret = cppc_get_perf_ctrs(i, &fb_ctrs);
+		if (ret) {
+			pr_warn("%s: failed to read perf counters: %d\n",
+				__func__, ret);
+			return cppc_freq_invariance_exit();
+		}
+
+		cppc_fi->prev_perf_fb_ctrs = fb_ctrs;
+	}
+
+	/* Register for freq-invariance */
+	topology_set_scale_freq_source(&cppc_sftd, cpu_present_mask);
+}
+
+#else
+static inline void
+cppc_freq_invariance_policy_init(struct cpufreq_policy *policy,
+				 struct cppc_cpudata *cpu_data)
+{
+}
+
+static inline void cppc_freq_invariance_exit(void)
+{
+}
+
+static inline void cppc_freq_invariance_init(void)
+{
+}
+#endif /* CONFIG_ACPI_CPPC_CPUFREQ_FIE */
+
 /* Callback function used to retrieve the max frequency from DMI */
 static void cppc_find_dmi_mhz(const struct dmi_header *dm, void *private)
 {
@@ -355,9 +556,12 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	cpu_data->perf_ctrls.desired_perf =  caps->highest_perf;
 
 	ret = cppc_set_perf(cpu, &cpu_data->perf_ctrls);
-	if (ret)
+	if (ret) {
 		pr_debug("Err setting perf value:%d on CPU:%d. ret:%d\n",
 			 caps->highest_perf, cpu, ret);
+	} else {
+		cppc_freq_invariance_policy_init(policy, cpu_data);
+	}
 
 	return ret;
 }
@@ -370,12 +574,12 @@ static inline u64 get_delta(u64 t1, u64 t0)
 	return (u32)t1 - (u32)t0;
 }
 
-static int cppc_get_rate_from_fbctrs(struct cppc_cpudata *cpu_data,
-				     struct cppc_perf_fb_ctrs fb_ctrs_t0,
-				     struct cppc_perf_fb_ctrs fb_ctrs_t1)
+static int cppc_perf_from_fbctrs(struct cppc_cpudata *cpu_data,
+				 struct cppc_perf_fb_ctrs fb_ctrs_t0,
+				 struct cppc_perf_fb_ctrs fb_ctrs_t1)
 {
 	u64 delta_reference, delta_delivered;
-	u64 reference_perf, delivered_perf;
+	u64 reference_perf;
 
 	reference_perf = fb_ctrs_t0.reference_perf;
 
@@ -384,12 +588,21 @@ static int cppc_get_rate_from_fbctrs(struct cppc_cpudata *cpu_data,
 	delta_delivered = get_delta(fb_ctrs_t1.delivered,
 				    fb_ctrs_t0.delivered);
 
-	/* Check to avoid divide-by zero */
-	if (delta_reference || delta_delivered)
-		delivered_perf = (reference_perf * delta_delivered) /
-					delta_reference;
-	else
-		delivered_perf = cpu_data->perf_ctrls.desired_perf;
+	/* Check to avoid divide-by zero and invalid delivered_perf */
+	if (!delta_reference || !delta_delivered)
+		return cpu_data->perf_ctrls.desired_perf;
+
+	return (reference_perf * delta_delivered) / delta_reference;
+}
+
+static int cppc_get_rate_from_fbctrs(struct cppc_cpudata *cpu_data,
+				     struct cppc_perf_fb_ctrs fb_ctrs_t0,
+				     struct cppc_perf_fb_ctrs fb_ctrs_t1)
+{
+	u64 delivered_perf;
+
+	delivered_perf = cppc_perf_from_fbctrs(cpu_data, fb_ctrs_t0,
+					       fb_ctrs_t1);
 
 	return cppc_cpufreq_perf_to_khz(cpu_data, delivered_perf);
 }
@@ -514,6 +727,8 @@ static void cppc_check_hisi_workaround(void)
 
 static int __init cppc_cpufreq_init(void)
 {
+	int ret;
+
 	if ((acpi_disabled) || !acpi_cpc_valid())
 		return -ENODEV;
 
@@ -521,7 +736,11 @@ static int __init cppc_cpufreq_init(void)
 
 	cppc_check_hisi_workaround();
 
-	return cpufreq_register_driver(&cppc_cpufreq_driver);
+	ret = cpufreq_register_driver(&cppc_cpufreq_driver);
+	if (!ret)
+		cppc_freq_invariance_init();
+
+	return ret;
 }
 
 static inline void free_cpu_data(void)
@@ -538,6 +757,7 @@ static inline void free_cpu_data(void)
 
 static void __exit cppc_cpufreq_exit(void)
 {
+	cppc_freq_invariance_exit();
 	cpufreq_unregister_driver(&cppc_cpufreq_driver);
 
 	free_cpu_data();
diff --git a/include/linux/arch_topology.h b/include/linux/arch_topology.h
index 3bcfba5c21a7..47ac4b41c28d 100644
--- a/include/linux/arch_topology.h
+++ b/include/linux/arch_topology.h
@@ -37,6 +37,7 @@ bool topology_scale_freq_invariant(void);
 enum scale_freq_source {
 	SCALE_FREQ_SOURCE_CPUFREQ = 0,
 	SCALE_FREQ_SOURCE_ARCH,
+	SCALE_FREQ_SOURCE_CPPC,
 };
 
 struct scale_freq_data {
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index ca2bb629595f..3adedc7b1725 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -6386,6 +6386,7 @@ int sched_setattr_nocheck(struct task_struct *p, const struct sched_attr *attr)
 {
 	return __sched_setscheduler(p, attr, false, true);
 }
+EXPORT_SYMBOL_GPL(sched_setattr_nocheck);
 
 /**
  * sched_setscheduler_nocheck - change the scheduling policy and/or RT priority of a thread from kernelspace.
-- 
2.25.0.rc1.19.g042ed3e048af

Re: [PATCH V5 2/2] cpufreq: CPPC: Add support for frequency invariance

[Ionela Voinescu]

Hey,

On Monday 01 Mar 2021 at 12:21:18 (+0530), Viresh Kumar wrote:
> The Frequency Invariance Engine (FIE) is providing a frequency scaling
> correction factor that helps achieve more accurate load-tracking.
> 
> Normally, this scaling factor can be obtained directly with the help of
> the cpufreq drivers as they know the exact frequency the hardware is
> running at. But that isn't the case for CPPC cpufreq driver.
> 
> Another way of obtaining that is using the arch specific counter
> support, which is already present in kernel, but that hardware is
> optional for platforms.
> 
> This patch updates the CPPC driver to register itself with the topology
> core to provide its own implementation (cppc_scale_freq_tick()) of
> topology_scale_freq_tick() which gets called by the scheduler on every
> tick. Note that the arch specific counters have higher priority than
> CPPC counters, if available, though the CPPC driver doesn't need to have
> any special handling for that.
> 
> On an invocation of cppc_scale_freq_tick(), we schedule an irq work
> (since we reach here from hard-irq context), which then schedules a
> normal work item and cppc_scale_freq_workfn() updates the per_cpu
> freq_scale variable based on the counter updates since the last tick.
> 
> To allow platforms to disable frequency invariance support if they want,
                               ^
			       this CPPC counter-based frequency invariance
			       support..

(disabling this config will not disable cpufreq or arch counter-based FIE)

> this is all done under CONFIG_ACPI_CPPC_CPUFREQ_FIE, which is enabled by
> default.
> 
> This also exports sched_setattr_nocheck() as the CPPC driver can be
> built as a module.
> 
> Cc: Ionela Voinescu <ionela.voinescu@arm.com>
> Cc: linux-acpi@vger.kernel.org
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> ---
>  drivers/cpufreq/Kconfig.arm    |   9 ++
>  drivers/cpufreq/cppc_cpufreq.c | 244 +++++++++++++++++++++++++++++++--
>  include/linux/arch_topology.h  |   1 +
>  kernel/sched/core.c            |   1 +
>  4 files changed, 243 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
> index e65e0a43be64..a3e2d6dfea70 100644
> --- a/drivers/cpufreq/Kconfig.arm
> +++ b/drivers/cpufreq/Kconfig.arm
> @@ -19,6 +19,15 @@ config ACPI_CPPC_CPUFREQ
>  
>  	  If in doubt, say N.
>  
> +config ACPI_CPPC_CPUFREQ_FIE
> +	bool "Frequency Invariance support for CPPC cpufreq driver"
> +	depends on ACPI_CPPC_CPUFREQ

It also depends on GENERIC_ARCH_TOPOLOGY.

> +	default y
> +	help
> +	  This enables frequency invariance support for CPPC cpufreq driver.
                                                    ^^^^^^^^^^^^^^^^^^^^^^^^
						    s//based on CPPC counters.

.. or more detailed: This extends frequency invariance support in the
CPPC cpufreq driver, by using CPPC delivered and reference performance
counters.

> +
> +	  If in doubt, say N.
> +
>  config ARM_ALLWINNER_SUN50I_CPUFREQ_NVMEM
>  	tristate "Allwinner nvmem based SUN50I CPUFreq driver"
>  	depends on ARCH_SUNXI
> diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
> index 8a482c434ea6..c4580a37a1b1 100644
> --- a/drivers/cpufreq/cppc_cpufreq.c
> +++ b/drivers/cpufreq/cppc_cpufreq.c
[..]
> +static void cppc_freq_invariance_policy_init(struct cpufreq_policy *policy,
> +					     struct cppc_cpudata *cpu_data)
> +{
> +	struct cppc_freq_invariance *cppc_fi;
> +	int i;
> +
> +	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> +		return;
> +
> +	for_each_cpu(i, policy->cpus) {
> +		cppc_fi = &per_cpu(cppc_freq_inv, i);
> +		cppc_fi->cpu = i;
> +		cppc_fi->cpu_data = cpu_data;
> +		kthread_init_work(&cppc_fi->work, cppc_scale_freq_workfn);
> +		init_irq_work(&cppc_fi->irq_work, cppc_irq_work);
> +	}
> +}
> +
> +static void cppc_freq_invariance_exit(void)
> +{
> +	struct cppc_freq_invariance *cppc_fi;
> +	int i;
> +
> +	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> +		return;
> +
> +	topology_clear_scale_freq_source(SCALE_FREQ_SOURCE_CPPC, cpu_present_mask);
> +
> +	for_each_possible_cpu(i) {
> +		cppc_fi = &per_cpu(cppc_freq_inv, i);
> +		irq_work_sync(&cppc_fi->irq_work);
> +	}
> +
> +	kthread_destroy_worker(kworker_fie);
> +	kworker_fie = NULL;
> +}
> +
> +static void __init cppc_freq_invariance_init(void)
> +{
> +	struct cppc_perf_fb_ctrs fb_ctrs = {0};
> +	struct cppc_freq_invariance *cppc_fi;
> +	struct sched_attr attr = {
> +		.size		= sizeof(struct sched_attr),
> +		.sched_policy	= SCHED_DEADLINE,
> +		.sched_nice	= 0,
> +		.sched_priority	= 0,
> +		/*
> +		 * Fake (unused) bandwidth; workaround to "fix"
> +		 * priority inheritance.
> +		 */
> +		.sched_runtime	= 1000000,
> +		.sched_deadline = 10000000,
> +		.sched_period	= 10000000,
> +	};
> +	int i, ret;
> +
> +	if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> +		return;
> +
> +	kworker_fie = kthread_create_worker(0, "cppc_fie");
> +	if (IS_ERR(kworker_fie))
> +		return;
> +
> +	ret = sched_setattr_nocheck(kworker_fie->task, &attr);
> +	if (ret) {
> +		pr_warn("%s: failed to set SCHED_DEADLINE: %d\n", __func__,
> +			ret);
> +		kthread_destroy_worker(kworker_fie);
> +		return;
> +	}
> +

Nit: to me it makes more sense to move the code below to 
cppc_freq_invariance_policy_init(). It seems a bit strange to do part of
the initialization of the per-cpu information there, and part here. But
I do understand the reasons for it. Moving the code below would also
save some cycles going through the CPUs again and will mimic the
frequency invariance setup process in the arm64 topology, where we do
amu_fie_setup() at policy creation time.

It's not a big deal so I'll leave it up to you.

> +	for_each_possible_cpu(i) {
> +		cppc_fi = &per_cpu(cppc_freq_inv, i);
> +
> +		/* A policy failed to initialize, abort */
> +		if (unlikely(!cppc_fi->cpu_data))
> +			return cppc_freq_invariance_exit();
> +
> +		ret = cppc_get_perf_ctrs(i, &fb_ctrs);
> +		if (ret) {
> +			pr_warn("%s: failed to read perf counters: %d\n",
> +				__func__, ret);
> +			return cppc_freq_invariance_exit();
> +		}
> +
> +		cppc_fi->prev_perf_fb_ctrs = fb_ctrs;
> +	}
> +
> +	/* Register for freq-invariance */
> +	topology_set_scale_freq_source(&cppc_sftd, cpu_present_mask);
> +}

After another very quick round of testing:

Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com>
Tested-by: Ionela Voinescu <ionela.voinescu@arm.com>

I did not get the chance to test on ThunderX2 yet, but if you are happy
with your testing on it, I won't delay this any further.

Thanks,
Ionela.

Re: [PATCH V5 2/2] cpufreq: CPPC: Add support for frequency invariance

[Vincent Guittot]

On Tue, 9 Mar 2021 at 16:10, Ionela Voinescu <ionela.voinescu@arm.com> wrote:
>
> Hey,
>
> On Monday 01 Mar 2021 at 12:21:18 (+0530), Viresh Kumar wrote:
> > The Frequency Invariance Engine (FIE) is providing a frequency scaling
> > correction factor that helps achieve more accurate load-tracking.
> >
> > Normally, this scaling factor can be obtained directly with the help of
> > the cpufreq drivers as they know the exact frequency the hardware is
> > running at. But that isn't the case for CPPC cpufreq driver.
> >
> > Another way of obtaining that is using the arch specific counter
> > support, which is already present in kernel, but that hardware is
> > optional for platforms.
> >
> > This patch updates the CPPC driver to register itself with the topology
> > core to provide its own implementation (cppc_scale_freq_tick()) of
> > topology_scale_freq_tick() which gets called by the scheduler on every
> > tick. Note that the arch specific counters have higher priority than
> > CPPC counters, if available, though the CPPC driver doesn't need to have
> > any special handling for that.
> >
> > On an invocation of cppc_scale_freq_tick(), we schedule an irq work
> > (since we reach here from hard-irq context), which then schedules a
> > normal work item and cppc_scale_freq_workfn() updates the per_cpu
> > freq_scale variable based on the counter updates since the last tick.
> >
> > To allow platforms to disable frequency invariance support if they want,
>                                ^
>                                this CPPC counter-based frequency invariance
>                                support..
>
> (disabling this config will not disable cpufreq or arch counter-based FIE)
>
> > this is all done under CONFIG_ACPI_CPPC_CPUFREQ_FIE, which is enabled by
> > default.
> >
> > This also exports sched_setattr_nocheck() as the CPPC driver can be
> > built as a module.
> >
> > Cc: Ionela Voinescu <ionela.voinescu@arm.com>
> > Cc: linux-acpi@vger.kernel.org
> > Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> > ---
> >  drivers/cpufreq/Kconfig.arm    |   9 ++
> >  drivers/cpufreq/cppc_cpufreq.c | 244 +++++++++++++++++++++++++++++++--
> >  include/linux/arch_topology.h  |   1 +
> >  kernel/sched/core.c            |   1 +
> >  4 files changed, 243 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
> > index e65e0a43be64..a3e2d6dfea70 100644
> > --- a/drivers/cpufreq/Kconfig.arm
> > +++ b/drivers/cpufreq/Kconfig.arm
> > @@ -19,6 +19,15 @@ config ACPI_CPPC_CPUFREQ
> >
> >         If in doubt, say N.
> >
> > +config ACPI_CPPC_CPUFREQ_FIE
> > +     bool "Frequency Invariance support for CPPC cpufreq driver"
> > +     depends on ACPI_CPPC_CPUFREQ
>
> It also depends on GENERIC_ARCH_TOPOLOGY.
>
> > +     default y
> > +     help
> > +       This enables frequency invariance support for CPPC cpufreq driver.
>                                                     ^^^^^^^^^^^^^^^^^^^^^^^^
>                                                     s//based on CPPC counters.
>
> .. or more detailed: This extends frequency invariance support in the
> CPPC cpufreq driver, by using CPPC delivered and reference performance
> counters.
>
> > +
> > +       If in doubt, say N.
> > +
> >  config ARM_ALLWINNER_SUN50I_CPUFREQ_NVMEM
> >       tristate "Allwinner nvmem based SUN50I CPUFreq driver"
> >       depends on ARCH_SUNXI
> > diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
> > index 8a482c434ea6..c4580a37a1b1 100644
> > --- a/drivers/cpufreq/cppc_cpufreq.c
> > +++ b/drivers/cpufreq/cppc_cpufreq.c
> [..]
> > +static void cppc_freq_invariance_policy_init(struct cpufreq_policy *policy,
> > +                                          struct cppc_cpudata *cpu_data)
> > +{
> > +     struct cppc_freq_invariance *cppc_fi;
> > +     int i;
> > +
> > +     if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> > +             return;
> > +
> > +     for_each_cpu(i, policy->cpus) {
> > +             cppc_fi = &per_cpu(cppc_freq_inv, i);
> > +             cppc_fi->cpu = i;
> > +             cppc_fi->cpu_data = cpu_data;
> > +             kthread_init_work(&cppc_fi->work, cppc_scale_freq_workfn);
> > +             init_irq_work(&cppc_fi->irq_work, cppc_irq_work);
> > +     }
> > +}
> > +
> > +static void cppc_freq_invariance_exit(void)
> > +{
> > +     struct cppc_freq_invariance *cppc_fi;
> > +     int i;
> > +
> > +     if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> > +             return;
> > +
> > +     topology_clear_scale_freq_source(SCALE_FREQ_SOURCE_CPPC, cpu_present_mask);
> > +
> > +     for_each_possible_cpu(i) {
> > +             cppc_fi = &per_cpu(cppc_freq_inv, i);
> > +             irq_work_sync(&cppc_fi->irq_work);
> > +     }
> > +
> > +     kthread_destroy_worker(kworker_fie);
> > +     kworker_fie = NULL;
> > +}
> > +
> > +static void __init cppc_freq_invariance_init(void)
> > +{
> > +     struct cppc_perf_fb_ctrs fb_ctrs = {0};
> > +     struct cppc_freq_invariance *cppc_fi;
> > +     struct sched_attr attr = {
> > +             .size           = sizeof(struct sched_attr),
> > +             .sched_policy   = SCHED_DEADLINE,
> > +             .sched_nice     = 0,
> > +             .sched_priority = 0,
> > +             /*
> > +              * Fake (unused) bandwidth; workaround to "fix"
> > +              * priority inheritance.
> > +              */
> > +             .sched_runtime  = 1000000,
> > +             .sched_deadline = 10000000,
> > +             .sched_period   = 10000000,
> > +     };
> > +     int i, ret;
> > +
> > +     if (cppc_cpufreq_driver.get == hisi_cppc_cpufreq_get_rate)
> > +             return;
> > +
> > +     kworker_fie = kthread_create_worker(0, "cppc_fie");
> > +     if (IS_ERR(kworker_fie))
> > +             return;
> > +
> > +     ret = sched_setattr_nocheck(kworker_fie->task, &attr);
> > +     if (ret) {
> > +             pr_warn("%s: failed to set SCHED_DEADLINE: %d\n", __func__,
> > +                     ret);
> > +             kthread_destroy_worker(kworker_fie);
> > +             return;
> > +     }
> > +
>
> Nit: to me it makes more sense to move the code below to
> cppc_freq_invariance_policy_init(). It seems a bit strange to do part of
> the initialization of the per-cpu information there, and part here. But
> I do understand the reasons for it. Moving the code below would also
> save some cycles going through the CPUs again and will mimic the
> frequency invariance setup process in the arm64 topology, where we do
> amu_fie_setup() at policy creation time.
>
> It's not a big deal so I'll leave it up to you.
>
> > +     for_each_possible_cpu(i) {
> > +             cppc_fi = &per_cpu(cppc_freq_inv, i);
> > +
> > +             /* A policy failed to initialize, abort */
> > +             if (unlikely(!cppc_fi->cpu_data))
> > +                     return cppc_freq_invariance_exit();
> > +
> > +             ret = cppc_get_perf_ctrs(i, &fb_ctrs);
> > +             if (ret) {
> > +                     pr_warn("%s: failed to read perf counters: %d\n",
> > +                             __func__, ret);
> > +                     return cppc_freq_invariance_exit();
> > +             }
> > +
> > +             cppc_fi->prev_perf_fb_ctrs = fb_ctrs;
> > +     }
> > +
> > +     /* Register for freq-invariance */
> > +     topology_set_scale_freq_source(&cppc_sftd, cpu_present_mask);
> > +}
>
> After another very quick round of testing:
>
> Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com>
> Tested-by: Ionela Voinescu <ionela.voinescu@arm.com>
>
> I did not get the chance to test on ThunderX2 yet, but if you are happy
> with your testing on it, I won't delay this any further.

I have just run some functional  tests on thx2 with rt-app: I have run
a periodic task (6ms running / 30ms periods) at different frequencies
(2.5Ghz, 2Ghz, 1.5Ghz, 1.333Ghz, 1Ghz) and the PELT signals stays the
same for all frequencies.

Tested-by: Vincent Guittot <vincent.guittot@linaro.org>


>
> Thanks,
> Ionela.