Merge tag 'pm-6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "The amd-pstate cpufreq driver gets the majority of changes this time.
  They are mostly fixes and cleanups, but one of them causes it to
  become the default cpufreq driver on some AMD server platforms.

  Apart from that, the menu cpuidle governor is modified to not use
  iowait any more, the intel_idle gets a custom C-states table for
  Granite Rapids Xeon D, and the intel_pstate driver will use a more
  aggressive Balance- performance default EPP value on Granite Rapids
  now.

  There are also some fixes, cleanups and tooling updates.

  Specifics:

   - Update the amd-pstate driver to set the initial scaling frequency
     policy lower bound to be the lowest non-linear frequency (Dhananjay
     Ugwekar)

   - Enable amd-pstate by default on servers starting with newer AMD
     Epyc processors (Swapnil Sapkal)

   - Align more codepaths between shared memory and MSR designs in
     amd-pstate (Dhananjay Ugwekar)

   - Clean up amd-pstate code to rename functions and remove redundant
     calls (Dhananjay Ugwekar, Mario Limonciello)

   - Do other assorted fixes and cleanups in amd-pstate (Dhananjay
     Ugwekar and Mario Limonciello)

   - Change the Balance-performance EPP value for Granite Rapids in the
     intel_pstate driver to a more performance-biased one (Srinivas
     Pandruvada)

   - Simplify MSR read on the boot CPU in the ACPI cpufreq driver (Chang
     S. Bae)

   - Ensure sugov_eas_rebuild_sd() is always called when sugov_init()
     succeeds to always enforce sched domains rebuild in case EAS needs
     to be enabled (Christian Loehle)

   - Switch cpufreq back to platform_driver::remove() (Uwe Kleine-König)

   - Use proper frequency unit names in cpufreq (Marcin Juszkiewicz)

   - Add a built-in idle states table for Granite Rapids Xeon D to the
     intel_idle driver (Artem Bityutskiy)

   - Fix some typos in comments in the cpuidle core and drivers (Shen
     Lichuan)

   - Remove iowait influence from the menu cpuidle governor (Christian
     Loehle)

   - Add min/max available performance state limits to the Energy Model
     management code (Lukasz Luba)

   - Update pm-graph to v5.13 (Todd Brandt)

   - Add documentation for some recently introduced cpupower utility
     options (Tor Vic)

   - Make cpupower inform users where cpufreq-bench.conf should be
     located when opening it fails (Peng Fan)

   - Allow overriding cross-compiling env params in cpupower (Peng Fan)

   - Add compile_commands.json to .gitignore in cpupower (John B. Wyatt
     IV)

   - Improve disable c_state block in cpupower bindings and add a test
     to confirm that CPU state is disabled to it (John B. Wyatt IV)

   - Add Chinese Simplified translation to cpupower (Kieran Moy)

   - Add checks for xgettext and msgfmt to cpupower (Siddharth Menon)"

* tag 'pm-6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (38 commits)
  cpufreq: intel_pstate: Update Balance-performance EPP for Granite Rapids
  cpufreq: ACPI: Simplify MSR read on the boot CPU
  sched/cpufreq: Ensure sd is rebuilt for EAS check
  intel_idle: add Granite Rapids Xeon D support
  PM: EM: Add min/max available performance state limits
  cpufreq/amd-pstate: Move registration after static function call update
  cpufreq/amd-pstate: Push adjust_perf vfunc init into cpu_init
  cpufreq/amd-pstate: Align offline flow of shared memory and MSR based systems
  cpufreq/amd-pstate: Call cppc_set_epp_perf in the reenable function
  cpufreq/amd-pstate: Do not attempt to clear MSR_AMD_CPPC_ENABLE
  cpufreq/amd-pstate: Rename functions that enable CPPC
  cpufreq/amd-pstate-ut: Add fix for min freq unit test
  amd-pstate: Switch to amd-pstate by default on some Server platforms
  amd-pstate: Set min_perf to nominal_perf for active mode performance gov
  cpufreq/amd-pstate: Remove the redundant amd_pstate_set_driver() call
  cpufreq/amd-pstate: Remove the switch case in amd_pstate_init()
  cpufreq/amd-pstate: Call amd_pstate_set_driver() in amd_pstate_register_driver()
  cpufreq/amd-pstate: Call amd_pstate_register() in amd_pstate_init()
  cpufreq/amd-pstate: Set the initial min_freq to lowest_nonlinear_freq
  cpufreq/amd-pstate: Remove the redundant verify() function
  ...

1 files changed, 21 insertions, 8 deletions

diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
index 1ff52020cf75..752e0b297582 100644
--- a/include/linux/energy_model.h
+++ b/include/linux/energy_model.h
@@ -55,6 +55,8 @@ struct em_perf_table {
  * struct em_perf_domain - Performance domain
  * @em_table:		Pointer to the runtime modifiable em_perf_table
  * @nr_perf_states:	Number of performance states
+ * @min_perf_state:	Minimum allowed Performance State index
+ * @max_perf_state:	Maximum allowed Performance State index
  * @flags:		See "em_perf_domain flags"
  * @cpus:		Cpumask covering the CPUs of the domain. It's here
  *			for performance reasons to avoid potential cache
@@ -70,6 +72,8 @@ struct em_perf_table {
 struct em_perf_domain {
 	struct em_perf_table __rcu *em_table;
 	int nr_perf_states;
+	int min_perf_state;
+	int max_perf_state;
 	unsigned long flags;
 	unsigned long cpus[];
 };
@@ -173,13 +177,14 @@ void em_table_free(struct em_perf_table __rcu *table);
 int em_dev_compute_costs(struct device *dev, struct em_perf_state *table,
 			 int nr_states);
 int em_dev_update_chip_binning(struct device *dev);
+int em_update_performance_limits(struct em_perf_domain *pd,
+		unsigned long freq_min_khz, unsigned long freq_max_khz);
 
 /**
  * em_pd_get_efficient_state() - Get an efficient performance state from the EM
  * @table:		List of performance states, in ascending order
- * @nr_perf_states:	Number of performance states
+ * @pd:			performance domain for which this must be done
  * @max_util:		Max utilization to map with the EM
- * @pd_flags:		Performance Domain flags
  *
  * It is called from the scheduler code quite frequently and as a consequence
  * doesn't implement any check.
@@ -188,13 +193,16 @@ int em_dev_update_chip_binning(struct device *dev);
  * requirement.
  */
 static inline int
-em_pd_get_efficient_state(struct em_perf_state *table, int nr_perf_states,
-			  unsigned long max_util, unsigned long pd_flags)
+em_pd_get_efficient_state(struct em_perf_state *table,
+			  struct em_perf_domain *pd, unsigned long max_util)
 {
+	unsigned long pd_flags = pd->flags;
+	int min_ps = pd->min_perf_state;
+	int max_ps = pd->max_perf_state;
 	struct em_perf_state *ps;
 	int i;
 
-	for (i = 0; i < nr_perf_states; i++) {
+	for (i = min_ps; i <= max_ps; i++) {
 		ps = &table[i];
 		if (ps->performance >= max_util) {
 			if (pd_flags & EM_PERF_DOMAIN_SKIP_INEFFICIENCIES &&
@@ -204,7 +212,7 @@ em_pd_get_efficient_state(struct em_perf_state *table, int nr_perf_states,
 		}
 	}
 
-	return nr_perf_states - 1;
+	return max_ps;
 }
 
 /**
@@ -253,8 +261,7 @@ static inline unsigned long em_cpu_energy(struct em_perf_domain *pd,
 	 * requested performance.
 	 */
 	em_table = rcu_dereference(pd->em_table);
-	i = em_pd_get_efficient_state(em_table->state, pd->nr_perf_states,
-				      max_util, pd->flags);
+	i = em_pd_get_efficient_state(em_table->state, pd, max_util);
 	ps = &em_table->state[i];
 
 	/*
@@ -391,6 +398,12 @@ static inline int em_dev_update_chip_binning(struct device *dev)
 {
 	return -EINVAL;
 }
+static inline
+int em_update_performance_limits(struct em_perf_domain *pd,
+		unsigned long freq_min_khz, unsigned long freq_max_khz)
+{
+	return -EINVAL;
+}
 #endif
 
 #endif