[PATCH RESEND v4 00/10] PM: EM: Add netlink support for the energy model

[PATCH RESEND v4 00/10] PM: EM: Add netlink support for the energy model

[Changwoo Min]

Rebased the code to current HEAD of the linus tree.

There is a need to access the energy model from the userspace. One such
example is the sched_ext schedulers [1]. The userspace part of the
sched_ext schedules could feed the (post-processed) energy-model
information to the BPF part of the scheduler.

Currently, debugfs is the only way to read the energy model from userspace;
however, it lacks proper notification mechanisms when a performance domain
and its associated energy model change.

This patch set introduces a generic netlink for the energy model, as
discussed in [2]. It allows a userspace program to read the performance
domain and its energy model. It notifies the userspace program when a
performance domain is created or deleted or its energy model is updated
through a multicast interface.

Specifically, it supports two commands:
  - EM_CMD_GET_PDS: Get the list of information for all performance
    domains.
  - EM_CMD_GET_PD_TABLE: Get the energy model table of a performance
    domain.

Also, it supports three notification events:
  - EM_CMD_PD_CREATED: When a performance domain is created.
  - EM_CMD_PD_DELETED: When a performance domain is deleted.
  - EM_CMD_PD_UPDATED: When the energy model table of a performance domain
    is updated.

This can be tested using the tool, tools/net/ynl/pyynl/cli.py, for example,
with the following commands:

  $> tools/net/ynl/pyynl/cli.py \
     --spec Documentation/netlink/specs/em.yaml \
     --do get-pds
  $> tools/net/ynl/pyynl/cli.py \
     --spec Documentation/netlink/specs/em.yaml \
     --do get-pd-table --json '{"pd-id": 0}'
  $> tools/net/ynl/pyynl/cli.py \
     --spec Documentation/netlink/specs/em.yaml \
     --subscribe event  --sleep 10

[1] https://lwn.net/Articles/922405/
[2] https://lore.kernel.org/lkml/a82423bc-8c38-4d57-93da-c4f20011cc92@arm.com/
[3] https://lore.kernel.org/lkml/202506140306.tuIoz8rN-lkp@intel.com/#t

ChangeLog v3 -> v4:
  - Move patches [3-5] to the first.
  - Remove the ending period (".") from all of the patch subjects.
  - Rebase the code to v6.17-rc4.

ChangeLog v2 -> v3:
  - Properly initialize a return variable in
    em_notify_pd_created/updated() at an error path (09/10), reported by
    the kernel test robot [3].
  - Remove redundant initialization of a return variable in
    em_notify_pd_deleted() at an error path (08/10).

ChangeLog v1 -> v2:
  - Use YNL to generate boilerplate code. Overhaul the naming conventions
    (command, event, notification, attribute) to follow the typical
    conventions of other YNL-based netlink implementations.
  - Calculate the exact message size instead of using NLMSG_GOODSIZE
    when allocating a message (genlmsg_new). This avoids the reallocation
    of a message.
  - Remove an unnecessary function, em_netlink_exit(), and initialize the
    netlink (em_netlink_init) at em_netlink.c without touching energy_model.c.

Changwoo Min (10):
  PM: EM: Assign a unique ID when creating a performance domain
  PM: EM: Expose the ID of a performance domain via debugfs
  PM: EM: Add an iterator and accessor for the performance domain
  PM: EM: Add em.yaml and autogen files
  PM: EM: Add a skeleton code for netlink notification
  PM: EM: Implement em_nl_get_pds_doit()
  PM: EM: Implement em_nl_get_pd_table_doit()
  PM: EM: Implement em_notify_pd_deleted()
  PM: EM: Implement em_notify_pd_created/updated()
  PM: EM: Notify an event when the performance domain changes

 Documentation/netlink/specs/em.yaml | 113 ++++++++++
 MAINTAINERS                         |   3 +
 include/linux/energy_model.h        |  19 ++
 include/uapi/linux/energy_model.h   |  62 ++++++
 kernel/power/Makefile               |   5 +-
 kernel/power/em_netlink.c           | 311 ++++++++++++++++++++++++++++
 kernel/power/em_netlink.h           |  34 +++
 kernel/power/em_netlink_autogen.c   |  48 +++++
 kernel/power/em_netlink_autogen.h   |  23 ++
 kernel/power/energy_model.c         |  86 +++++++-
 10 files changed, 702 insertions(+), 2 deletions(-)
 create mode 100644 Documentation/netlink/specs/em.yaml
 create mode 100644 include/uapi/linux/energy_model.h
 create mode 100644 kernel/power/em_netlink.c
 create mode 100644 kernel/power/em_netlink.h
 create mode 100644 kernel/power/em_netlink_autogen.c
 create mode 100644 kernel/power/em_netlink_autogen.h

-- 
2.51.0

[PATCH RESEND v4 01/10] PM: EM: Assign a unique ID when creating a performance domain

[Changwoo Min]

It is necessary to refer to a specific performance domain from a
userspace. For example, the energy model of a particular performance
domain is updated.

To this end, assign a unique ID to each performance domain to address it,
and manage them in a global linked list to look up a specific one by
matching ID. IDA is used for ID assignment, and the mutex is used to
protect the global list from concurrent access.

Note that the mutex (em_pd_list_mutex) is not supposed to hold while
holding em_pd_mutex to avoid ABBA deadlock.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 include/linux/energy_model.h |  4 ++++
 kernel/power/energy_model.c  | 33 ++++++++++++++++++++++++++++++++-
 2 files changed, 36 insertions(+), 1 deletion(-)

diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
index 61d50571ad88..43aa6153dc57 100644
--- a/include/linux/energy_model.h
+++ b/include/linux/energy_model.h
@@ -54,6 +54,8 @@ struct em_perf_table {
 /**
  * struct em_perf_domain - Performance domain
  * @em_table:		Pointer to the runtime modifiable em_perf_table
+ * @node:		node in	em_pd_list (in energy_model.c)
+ * @id:			A unique ID number for each performance domain
  * @nr_perf_states:	Number of performance states
  * @min_perf_state:	Minimum allowed Performance State index
  * @max_perf_state:	Maximum allowed Performance State index
@@ -71,6 +73,8 @@ struct em_perf_table {
  */
 struct em_perf_domain {
 	struct em_perf_table __rcu *em_table;
+	struct list_head node;
+	int id;
 	int nr_perf_states;
 	int min_perf_state;
 	int max_perf_state;
diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
index 8df55397414a..3fe562b6230e 100644
--- a/kernel/power/energy_model.c
+++ b/kernel/power/energy_model.c
@@ -23,6 +23,16 @@
  */
 static DEFINE_MUTEX(em_pd_mutex);
 
+/*
+ * Manage performance domains with IDs. One can iterate the performance domains
+ * through the list and pick one with their associated ID. The mutex serializes
+ * the list access. When holding em_pd_list_mutex, em_pd_mutex should not be
+ * taken to avoid potential deadlock.
+ */
+static DEFINE_IDA(em_pd_ida);
+static LIST_HEAD(em_pd_list);
+static DEFINE_MUTEX(em_pd_list_mutex);
+
 static void em_cpufreq_update_efficiencies(struct device *dev,
 					   struct em_perf_state *table);
 static void em_check_capacity_update(void);
@@ -396,7 +406,7 @@ static int em_create_pd(struct device *dev, int nr_states,
 	struct em_perf_table *em_table;
 	struct em_perf_domain *pd;
 	struct device *cpu_dev;
-	int cpu, ret, num_cpus;
+	int cpu, ret, num_cpus, id;
 
 	if (_is_cpu_device(dev)) {
 		num_cpus = cpumask_weight(cpus);
@@ -420,6 +430,13 @@ static int em_create_pd(struct device *dev, int nr_states,
 
 	pd->nr_perf_states = nr_states;
 
+	INIT_LIST_HEAD(&pd->node);
+
+	id = ida_alloc(&em_pd_ida, GFP_KERNEL);
+	if (id < 0)
+		return -ENOMEM;
+	pd->id = id;
+
 	em_table = em_table_alloc(pd);
 	if (!em_table)
 		goto free_pd;
@@ -444,6 +461,7 @@ static int em_create_pd(struct device *dev, int nr_states,
 	kfree(em_table);
 free_pd:
 	kfree(pd);
+	ida_free(&em_pd_ida, id);
 	return -EINVAL;
 }
 
@@ -660,6 +678,13 @@ int em_dev_register_pd_no_update(struct device *dev, unsigned int nr_states,
 unlock:
 	mutex_unlock(&em_pd_mutex);
 
+	if (_is_cpu_device(dev))
+		em_check_capacity_update();
+
+	mutex_lock(&em_pd_list_mutex);
+	list_add_tail(&dev->em_pd->node, &em_pd_list);
+	mutex_unlock(&em_pd_list_mutex);
+
 	return ret;
 }
 EXPORT_SYMBOL_GPL(em_dev_register_pd_no_update);
@@ -678,6 +703,10 @@ void em_dev_unregister_perf_domain(struct device *dev)
 	if (_is_cpu_device(dev))
 		return;
 
+	mutex_lock(&em_pd_list_mutex);
+	list_del_init(&dev->em_pd->node);
+	mutex_unlock(&em_pd_list_mutex);
+
 	/*
 	 * The mutex separates all register/unregister requests and protects
 	 * from potential clean-up/setup issues in the debugfs directories.
@@ -689,6 +718,8 @@ void em_dev_unregister_perf_domain(struct device *dev)
 	em_table_free(rcu_dereference_protected(dev->em_pd->em_table,
 						lockdep_is_held(&em_pd_mutex)));
 
+	ida_free(&em_pd_ida, dev->em_pd->id);
+
 	kfree(dev->em_pd);
 	dev->em_pd = NULL;
 	mutex_unlock(&em_pd_mutex);
-- 
2.51.0

[PATCH RESEND v4 02/10] PM: EM: Expose the ID of a performance domain via debugfs

[Changwoo Min]

For ease of debugging, let's expose the assigned ID of a performance
domain through debugfs (e.g., /sys/kernel/debug/energy_model/cpu0/id).

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/energy_model.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
index 3fe562b6230e..8998a7f4910a 100644
--- a/kernel/power/energy_model.c
+++ b/kernel/power/energy_model.c
@@ -126,6 +126,16 @@ static int em_debug_flags_show(struct seq_file *s, void *unused)
 }
 DEFINE_SHOW_ATTRIBUTE(em_debug_flags);
 
+static int em_debug_id_show(struct seq_file *s, void *unused)
+{
+	struct em_perf_domain *pd = s->private;
+
+	seq_printf(s, "%d\n", pd->id);
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(em_debug_id);
+
 static void em_debug_create_pd(struct device *dev)
 {
 	struct em_dbg_info *em_dbg;
@@ -142,6 +152,8 @@ static void em_debug_create_pd(struct device *dev)
 	debugfs_create_file("flags", 0444, d, dev->em_pd,
 			    &em_debug_flags_fops);
 
+	debugfs_create_file("id", 0444, d, dev->em_pd, &em_debug_id_fops);
+
 	em_dbg = devm_kcalloc(dev, dev->em_pd->nr_perf_states,
 			      sizeof(*em_dbg), GFP_KERNEL);
 	if (!em_dbg)
-- 
2.51.0

[PATCH RESEND v4 03/10] PM: EM: Add an iterator and accessor for the performance domain

[Changwoo Min]

Add an iterator function (for_each_em_perf_domain) that iterates all the
performance domains in the global list. A passed callback function (cb) is
called for each performance domain.

Additionally, add a lookup function (em_perf_domain_get_by_id) that
searches for a performance domain by matching the ID in the global list.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 include/linux/energy_model.h | 15 +++++++++++++++
 kernel/power/energy_model.c  | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+)

diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
index 43aa6153dc57..21279e779188 100644
--- a/include/linux/energy_model.h
+++ b/include/linux/energy_model.h
@@ -344,6 +344,10 @@ struct em_perf_state *em_perf_state_from_pd(struct em_perf_domain *pd)
 	return rcu_dereference(pd->em_table)->state;
 }
 
+int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
+			    void *data);
+struct em_perf_domain *em_perf_domain_get_by_id(int id);
+
 #else
 struct em_data_callback {};
 #define EM_ADV_DATA_CB(_active_power_cb, _cost_cb) { }
@@ -420,6 +424,17 @@ int em_update_performance_limits(struct em_perf_domain *pd,
 }
 static inline void em_adjust_cpu_capacity(unsigned int cpu) {}
 static inline void em_rebuild_sched_domains(void) {}
+static inline
+int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
+			    void *data)
+{
+	return -EINVAL;
+}
+static inline
+struct em_perf_domain *em_perf_domain_get_by_id(int id)
+{
+	return NULL;
+}
 #endif
 
 #endif
diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
index 8998a7f4910a..740076d24479 100644
--- a/kernel/power/energy_model.c
+++ b/kernel/power/energy_model.c
@@ -1000,3 +1000,37 @@ void em_rebuild_sched_domains(void)
 	 */
 	schedule_work(&rebuild_sd_work);
 }
+
+int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
+			    void *data)
+{
+	struct em_perf_domain *pd;
+
+	lockdep_assert_not_held(&em_pd_mutex);
+	guard(mutex)(&em_pd_list_mutex);
+
+	list_for_each_entry(pd, &em_pd_list, node) {
+		int ret;
+
+		ret = cb(pd, data);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+struct em_perf_domain *em_perf_domain_get_by_id(int id)
+{
+	struct em_perf_domain *pd;
+
+	lockdep_assert_not_held(&em_pd_mutex);
+	guard(mutex)(&em_pd_list_mutex);
+
+	list_for_each_entry(pd, &em_pd_list, node) {
+		if (pd->id == id)
+			return pd;
+	}
+
+	return NULL;
+}
-- 
2.51.0

[PATCH RESEND v4 04/10] PM: EM: Add em.yaml and autogen files

[Changwoo Min]

Add a generic netlink spec in YAML format and autogenerate boilerplate
code using ynl-regen.sh to introduce a generic netlink for the energy
model. It allows a userspace program to read the performance domain and
its energy model. It notifies the userspace program when a performance
domain is created or deleted or its energy model is updated through a
multicast interface.

Specifically, it supports two commands:
  - EM_CMD_GET_PDS: Get the list of information for all performance
    domains.
  - EM_CMD_GET_PD_TABLE: Get the energy model table of a performance
    domain.

Also, it supports three notification events:
  - EM_CMD_PD_CREATED: When a performance domain is created.
  - EM_CMD_PD_DELETED: When a performance domain is deleted.
  - EM_CMD_PD_UPDATED: When the energy model table of a performance domain
    is updated.

Finally, update MAINTAINERS to include new files.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 Documentation/netlink/specs/em.yaml | 113 ++++++++++++++++++++++++++++
 MAINTAINERS                         |   3 +
 include/uapi/linux/energy_model.h   |  62 +++++++++++++++
 kernel/power/em_netlink_autogen.c   |  48 ++++++++++++
 kernel/power/em_netlink_autogen.h   |  23 ++++++
 5 files changed, 249 insertions(+)
 create mode 100644 Documentation/netlink/specs/em.yaml
 create mode 100644 include/uapi/linux/energy_model.h
 create mode 100644 kernel/power/em_netlink_autogen.c
 create mode 100644 kernel/power/em_netlink_autogen.h

diff --git a/Documentation/netlink/specs/em.yaml b/Documentation/netlink/specs/em.yaml
new file mode 100644
index 000000000000..9905ca482325
--- /dev/null
+++ b/Documentation/netlink/specs/em.yaml
@@ -0,0 +1,113 @@
+# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+
+name: em
+
+doc: |
+  Energy model netlink interface to notify its changes.
+
+protocol: genetlink
+
+uapi-header: linux/energy_model.h
+
+attribute-sets:
+  -
+    name: pds
+    attributes:
+      -
+        name: pd
+        type: nest
+        nested-attributes: pd
+        multi-attr: true
+  -
+    name: pd
+    attributes:
+      -
+        name: pad
+        type: pad
+      -
+        name: pd-id
+        type: u32
+      -
+        name: flags
+        type: u64
+      -
+        name: cpus
+        type: string
+  -
+    name: pd-table
+    attributes:
+      -
+        name: pd-id
+        type: u32
+      -
+        name: ps
+        type: nest
+        nested-attributes: ps
+        multi-attr: true
+  -
+    name: ps
+    attributes:
+      -
+        name: pad
+        type: pad
+      -
+        name: performance
+        type: u64
+      -
+        name: frequency
+        type: u64
+      -
+        name: power
+        type: u64
+      -
+        name: cost
+        type: u64
+      -
+        name: flags
+        type: u64
+
+operations:
+  list:
+    -
+      name: get-pds
+      attribute-set: pds
+      doc: Get the list of information for all performance domains.
+      do:
+        reply:
+          attributes:
+            - pd
+    -
+      name: get-pd-table
+      attribute-set: pd-table
+      doc: Get the energy model table of a performance domain.
+      do:
+        request:
+          attributes:
+            - pd-id
+        reply:
+          attributes:
+            - pd-id
+            - ps
+    -
+      name: pd-created
+      doc: A performance domain is created.
+      notify: get-pd-table
+      mcgrp: event
+    -
+      name: pd-updated
+      doc: A performance domain is updated.
+      notify: get-pd-table
+      mcgrp: event
+    -
+      name: pd-deleted
+      doc: A performance domain is deleted.
+      attribute-set: pd-table
+      event:
+        attributes:
+            - pd-id
+      mcgrp: event
+
+mcast-groups:
+  list:
+    -
+      name: event
diff --git a/MAINTAINERS b/MAINTAINERS
index 520fb4e379a3..0992029d271d 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -9032,6 +9032,9 @@ S:	Maintained
 F:	kernel/power/energy_model.c
 F:	include/linux/energy_model.h
 F:	Documentation/power/energy-model.rst
+F:	Documentation/netlink/specs/em.yaml
+F:	include/uapi/linux/energy_model.h
+F:	kernel/power/em_netlink_autogen.*
 
 EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
 M:	Laurentiu Tudor <laurentiu.tudor@nxp.com>
diff --git a/include/uapi/linux/energy_model.h b/include/uapi/linux/energy_model.h
new file mode 100644
index 000000000000..4ec4c0eabbbb
--- /dev/null
+++ b/include/uapi/linux/energy_model.h
@@ -0,0 +1,62 @@
+/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
+/* Do not edit directly, auto-generated from: */
+/*	Documentation/netlink/specs/em.yaml */
+/* YNL-GEN uapi header */
+
+#ifndef _UAPI_LINUX_ENERGY_MODEL_H
+#define _UAPI_LINUX_ENERGY_MODEL_H
+
+#define EM_FAMILY_NAME		"em"
+#define EM_FAMILY_VERSION	1
+
+enum {
+	EM_A_PDS_PD = 1,
+
+	__EM_A_PDS_MAX,
+	EM_A_PDS_MAX = (__EM_A_PDS_MAX - 1)
+};
+
+enum {
+	EM_A_PD_PAD = 1,
+	EM_A_PD_PD_ID,
+	EM_A_PD_FLAGS,
+	EM_A_PD_CPUS,
+
+	__EM_A_PD_MAX,
+	EM_A_PD_MAX = (__EM_A_PD_MAX - 1)
+};
+
+enum {
+	EM_A_PD_TABLE_PD_ID = 1,
+	EM_A_PD_TABLE_PS,
+
+	__EM_A_PD_TABLE_MAX,
+	EM_A_PD_TABLE_MAX = (__EM_A_PD_TABLE_MAX - 1)
+};
+
+enum {
+	EM_A_PS_PAD = 1,
+	EM_A_PS_PERFORMANCE,
+	EM_A_PS_FREQUENCY,
+	EM_A_PS_POWER,
+	EM_A_PS_COST,
+	EM_A_PS_FLAGS,
+
+	__EM_A_PS_MAX,
+	EM_A_PS_MAX = (__EM_A_PS_MAX - 1)
+};
+
+enum {
+	EM_CMD_GET_PDS = 1,
+	EM_CMD_GET_PD_TABLE,
+	EM_CMD_PD_CREATED,
+	EM_CMD_PD_UPDATED,
+	EM_CMD_PD_DELETED,
+
+	__EM_CMD_MAX,
+	EM_CMD_MAX = (__EM_CMD_MAX - 1)
+};
+
+#define EM_MCGRP_EVENT	"event"
+
+#endif /* _UAPI_LINUX_ENERGY_MODEL_H */
diff --git a/kernel/power/em_netlink_autogen.c b/kernel/power/em_netlink_autogen.c
new file mode 100644
index 000000000000..a7a09ab1d1c2
--- /dev/null
+++ b/kernel/power/em_netlink_autogen.c
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+/* Do not edit directly, auto-generated from: */
+/*	Documentation/netlink/specs/em.yaml */
+/* YNL-GEN kernel source */
+
+#include <net/netlink.h>
+#include <net/genetlink.h>
+
+#include "em_netlink_autogen.h"
+
+#include <uapi/linux/energy_model.h>
+
+/* EM_CMD_GET_PD_TABLE - do */
+static const struct nla_policy em_get_pd_table_nl_policy[EM_A_PD_TABLE_PD_ID + 1] = {
+	[EM_A_PD_TABLE_PD_ID] = { .type = NLA_U32, },
+};
+
+/* Ops table for em */
+static const struct genl_split_ops em_nl_ops[] = {
+	{
+		.cmd	= EM_CMD_GET_PDS,
+		.doit	= em_nl_get_pds_doit,
+		.flags	= GENL_CMD_CAP_DO,
+	},
+	{
+		.cmd		= EM_CMD_GET_PD_TABLE,
+		.doit		= em_nl_get_pd_table_doit,
+		.policy		= em_get_pd_table_nl_policy,
+		.maxattr	= EM_A_PD_TABLE_PD_ID,
+		.flags		= GENL_CMD_CAP_DO,
+	},
+};
+
+static const struct genl_multicast_group em_nl_mcgrps[] = {
+	[EM_NLGRP_EVENT] = { "event", },
+};
+
+struct genl_family em_nl_family __ro_after_init = {
+	.name		= EM_FAMILY_NAME,
+	.version	= EM_FAMILY_VERSION,
+	.netnsok	= true,
+	.parallel_ops	= true,
+	.module		= THIS_MODULE,
+	.split_ops	= em_nl_ops,
+	.n_split_ops	= ARRAY_SIZE(em_nl_ops),
+	.mcgrps		= em_nl_mcgrps,
+	.n_mcgrps	= ARRAY_SIZE(em_nl_mcgrps),
+};
diff --git a/kernel/power/em_netlink_autogen.h b/kernel/power/em_netlink_autogen.h
new file mode 100644
index 000000000000..78ce609641f1
--- /dev/null
+++ b/kernel/power/em_netlink_autogen.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
+/* Do not edit directly, auto-generated from: */
+/*	Documentation/netlink/specs/em.yaml */
+/* YNL-GEN kernel header */
+
+#ifndef _LINUX_EM_GEN_H
+#define _LINUX_EM_GEN_H
+
+#include <net/netlink.h>
+#include <net/genetlink.h>
+
+#include <uapi/linux/energy_model.h>
+
+int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info);
+int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info);
+
+enum {
+	EM_NLGRP_EVENT,
+};
+
+extern struct genl_family em_nl_family;
+
+#endif /* _LINUX_EM_GEN_H */
-- 
2.51.0

[PATCH RESEND v4 05/10] PM: EM: Add a skeleton code for netlink notification

[Changwoo Min]

Add a boilerplate code for netlink notification to register the new
protocol family. Also, initialize and register the netlink during booting.
The initialization is called at the postcore level, which is late enough
after the generic netlink is initialized.

Finally, update MAINTAINERS to include new files.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 MAINTAINERS               |  2 +-
 kernel/power/Makefile     |  5 ++++-
 kernel/power/em_netlink.c | 35 +++++++++++++++++++++++++++++++++++
 kernel/power/em_netlink.h | 16 ++++++++++++++++
 4 files changed, 56 insertions(+), 2 deletions(-)
 create mode 100644 kernel/power/em_netlink.c
 create mode 100644 kernel/power/em_netlink.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 0992029d271d..ba528836eac1 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -9034,7 +9034,7 @@ F:	include/linux/energy_model.h
 F:	Documentation/power/energy-model.rst
 F:	Documentation/netlink/specs/em.yaml
 F:	include/uapi/linux/energy_model.h
-F:	kernel/power/em_netlink_autogen.*
+F:	kernel/power/em_netlink*.*
 
 EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
 M:	Laurentiu Tudor <laurentiu.tudor@nxp.com>
diff --git a/kernel/power/Makefile b/kernel/power/Makefile
index 874ad834dc8d..284a760aade7 100644
--- a/kernel/power/Makefile
+++ b/kernel/power/Makefile
@@ -21,4 +21,7 @@ obj-$(CONFIG_PM_WAKELOCKS)	+= wakelock.o
 
 obj-$(CONFIG_MAGIC_SYSRQ)	+= poweroff.o
 
-obj-$(CONFIG_ENERGY_MODEL)	+= energy_model.o
+obj-$(CONFIG_ENERGY_MODEL)	+= em.o
+em-y				:= energy_model.o
+em-$(CONFIG_NET)		+= em_netlink_autogen.o em_netlink.o
+
diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
new file mode 100644
index 000000000000..f3fbfeff29a4
--- /dev/null
+++ b/kernel/power/em_netlink.c
@@ -0,0 +1,35 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *
+ * Generic netlink for energy model.
+ *
+ * Copyright (c) 2025 Valve Corporation.
+ * Author: Changwoo Min <changwoo@igalia.com>
+ */
+
+#define pr_fmt(fmt) "energy_model: " fmt
+
+#include <linux/energy_model.h>
+#include <net/sock.h>
+#include <net/genetlink.h>
+#include <uapi/linux/energy_model.h>
+
+#include "em_netlink.h"
+#include "em_netlink_autogen.h"
+
+int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
+{
+	return -EOPNOTSUPP;
+}
+
+int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
+{
+	return -EOPNOTSUPP;
+}
+
+static int __init em_netlink_init(void)
+{
+	return genl_register_family(&em_nl_family);
+}
+postcore_initcall(em_netlink_init);
+
diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
new file mode 100644
index 000000000000..acd186c92d6b
--- /dev/null
+++ b/kernel/power/em_netlink.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ *
+ * Generic netlink for energy model.
+ *
+ * Copyright (c) 2025 Valve Corporation.
+ * Author: Changwoo Min <changwoo@igalia.com>
+ */
+#ifndef _EM_NETLINK_H
+#define _EM_NETLINK_H
+
+#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
+#else
+#endif
+
+#endif /* _EM_NETLINK_H */
-- 
2.51.0

[PATCH RESEND v4 06/10] PM: EM: Implement em_nl_get_pds_doit()

[Changwoo Min]

When a userspace requests EM_CMD_GET_PDS, the kernel responds with
information on all performance domains. The message format of the
response is as follows:

EM_A_PDS_PD (NLA_NESTED)*
    EM_A_PD_PD_ID (NLA_U32)
    EM_A_PD_FLAGS (NLA_U64)
    EM_A_PD_CPUS (NLA_STRING)

where EM_A_PDS_PD can be repeated as many times as there are performance
domains, and EM_A_PD_CPUS is a hexadecimal string representing a CPU
bitmask.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/em_netlink.c | 82 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 81 insertions(+), 1 deletion(-)

diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
index f3fbfeff29a4..31b27c6fe3c9 100644
--- a/kernel/power/em_netlink.c
+++ b/kernel/power/em_netlink.c
@@ -17,9 +17,89 @@
 #include "em_netlink.h"
 #include "em_netlink_autogen.h"
 
+#define EM_A_PD_CPUS_LEN		256
+
+/*************************** Command encoding ********************************/
+static int __em_nl_get_pd_size(struct em_perf_domain *pd, void *data)
+{
+	char cpus_buf[EM_A_PD_CPUS_LEN];
+	int *tot_msg_sz = data;
+	int msg_sz, cpus_sz;
+
+	cpus_sz = snprintf(cpus_buf, sizeof(cpus_buf), "%*pb",
+			   cpumask_pr_args(to_cpumask(pd->cpus)));
+
+	msg_sz = nla_total_size(0) +			/* EM_A_PDS_PD */
+		 nla_total_size(sizeof(u32)) +		/* EM_A_PD_PD_ID */
+		 nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PD_FLAGS */
+		 nla_total_size(cpus_sz);		/* EM_A_PD_CPUS */
+
+	*tot_msg_sz += nlmsg_total_size(genlmsg_msg_size(msg_sz));
+	return 0;
+}
+
+static int __em_nl_get_pd(struct em_perf_domain *pd, void *data)
+{
+	char cpus_buf[EM_A_PD_CPUS_LEN];
+	struct sk_buff *msg = data;
+	struct nlattr *entry;
+
+	entry = nla_nest_start(msg, EM_A_PDS_PD);
+	if (!entry)
+		goto out_cancel_nest;
+
+	if (nla_put_u32(msg, EM_A_PD_PD_ID, pd->id))
+		goto out_cancel_nest;
+
+	if (nla_put_u64_64bit(msg, EM_A_PD_FLAGS, pd->flags, EM_A_PD_PAD))
+		goto out_cancel_nest;
+
+	snprintf(cpus_buf, sizeof(cpus_buf), "%*pb",
+		 cpumask_pr_args(to_cpumask(pd->cpus)));
+	if (nla_put_string(msg, EM_A_PD_CPUS, cpus_buf))
+		goto out_cancel_nest;
+
+	nla_nest_end(msg, entry);
+
+	return 0;
+
+out_cancel_nest:
+	nla_nest_cancel(msg, entry);
+
+	return -EMSGSIZE;
+}
+
 int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
 {
-	return -EOPNOTSUPP;
+	struct sk_buff *msg;
+	void *hdr;
+	int cmd = info->genlhdr->cmd;
+	int ret = -EMSGSIZE, msg_sz = 0;
+
+	for_each_em_perf_domain(__em_nl_get_pd_size, &msg_sz);
+
+	msg = genlmsg_new(msg_sz, GFP_KERNEL);
+	if (!msg)
+		return -ENOMEM;
+
+	hdr = genlmsg_put_reply(msg, info, &em_nl_family, 0, cmd);
+	if (!hdr)
+		goto out_free_msg;
+
+	ret = for_each_em_perf_domain(__em_nl_get_pd, msg);
+	if (ret)
+		goto out_cancel_msg;
+
+	genlmsg_end(msg, hdr);
+
+	return genlmsg_reply(msg, info);
+
+out_cancel_msg:
+	genlmsg_cancel(msg, hdr);
+out_free_msg:
+	nlmsg_free(msg);
+
+	return ret;
 }
 
 int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
-- 
2.51.0

[PATCH RESEND v4 07/10] PM: EM: Implement em_nl_get_pd_table_doit()

[Changwoo Min]

When a userspace requests EM_CMD_GET_PD_TABLE with an ID of a performancei
domain, the kernel reports back the energy model table of the specified
performance domain. The message format of the response is as follows:

EM_A_PD_TABLE_PD_ID (NLA_U32)
EM_A_PD_TABLE_PS (NLA_NESTED)*
    EM_A_PS_PERFORMANCE (NLA_U64)
    EM_A_PS_FREQUENCY (NLA_U64)
    EM_A_PS_POWER (NLA_U64)
    EM_A_PS_COST (NLA_U64)
    EM_A_PS_FLAGS (NLA_U64)

where EM_A_PD_TABLE_PS can be repeated as many times as there are
performance states (struct em_perf_state).

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/em_netlink.c | 108 +++++++++++++++++++++++++++++++++++++-
 1 file changed, 107 insertions(+), 1 deletion(-)

diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
index 31b27c6fe3c9..59953cfedf78 100644
--- a/kernel/power/em_netlink.c
+++ b/kernel/power/em_netlink.c
@@ -102,9 +102,115 @@ int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
 	return ret;
 }
 
+static struct em_perf_domain *__em_nl_get_pd_table_id(struct nlattr **attrs)
+{
+	struct em_perf_domain *pd;
+	int id;
+
+	if (!attrs[EM_A_PD_TABLE_PD_ID])
+		return NULL;
+
+	id = nla_get_u32(attrs[EM_A_PD_TABLE_PD_ID]);
+	pd = em_perf_domain_get_by_id(id);
+	return pd;
+}
+
+static int __em_nl_get_pd_table_size(const struct em_perf_domain *pd)
+{
+	int id_sz, ps_sz;
+
+	id_sz = nla_total_size(sizeof(u32));		/* EM_A_PD_TABLE_PD_ID */
+	ps_sz = nla_total_size(0) +			/* EM_A_PD_TABLE_PS */
+		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_PERFORMANCE */
+		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_FREQUENCY */
+		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_POWER */
+		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_COST */
+		nla_total_size_64bit(sizeof(u64));	/* EM_A_PS_FLAGS */
+	ps_sz *= pd->nr_perf_states;
+
+	return nlmsg_total_size(genlmsg_msg_size(id_sz + ps_sz));
+}
+
+static int __em_nl_get_pd_table(struct sk_buff *msg, const struct em_perf_domain *pd)
+{
+	struct em_perf_state *table, *ps;
+	struct nlattr *entry;
+	int i;
+
+	if (nla_put_u32(msg, EM_A_PD_TABLE_PD_ID, pd->id))
+		goto out_err;
+
+	rcu_read_lock();
+	table = em_perf_state_from_pd((struct em_perf_domain *)pd);
+
+	for (i = 0; i < pd->nr_perf_states; i++) {
+		ps = &table[i];
+
+		entry = nla_nest_start(msg, EM_A_PD_TABLE_PS);
+		if (!entry)
+			goto out_unlock_ps;
+
+		if (nla_put_u64_64bit(msg, EM_A_PS_PERFORMANCE,
+				      ps->performance, EM_A_PS_PAD))
+			goto out_cancel_ps_nest;
+		if (nla_put_u64_64bit(msg, EM_A_PS_FREQUENCY,
+				      ps->frequency, EM_A_PS_PAD))
+			goto out_cancel_ps_nest;
+		if (nla_put_u64_64bit(msg, EM_A_PS_POWER,
+				      ps->power, EM_A_PS_PAD))
+			goto out_cancel_ps_nest;
+		if (nla_put_u64_64bit(msg, EM_A_PS_COST,
+				      ps->cost, EM_A_PS_PAD))
+			goto out_cancel_ps_nest;
+		if (nla_put_u64_64bit(msg, EM_A_PS_FLAGS,
+				      ps->flags, EM_A_PS_PAD))
+			goto out_cancel_ps_nest;
+
+		nla_nest_end(msg, entry);
+	}
+	rcu_read_unlock();
+	return 0;
+
+out_cancel_ps_nest:
+	nla_nest_cancel(msg, entry);
+out_unlock_ps:
+	rcu_read_unlock();
+out_err:
+	return -EMSGSIZE;
+}
+
 int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
 {
-	return -EOPNOTSUPP;
+	struct sk_buff *msg;
+	struct em_perf_domain *pd;
+	void *hdr;
+	int cmd = info->genlhdr->cmd;
+	int msg_sz, ret = -EMSGSIZE;
+
+	pd = __em_nl_get_pd_table_id(info->attrs);
+	if (!pd)
+		return -EINVAL;
+
+	msg_sz = __em_nl_get_pd_table_size(pd);
+
+	msg = genlmsg_new(msg_sz, GFP_KERNEL);
+	if (!msg)
+		return -ENOMEM;
+
+	hdr = genlmsg_put_reply(msg, info, &em_nl_family, 0, cmd);
+	if (!hdr)
+		goto out_free_msg;
+
+	ret = __em_nl_get_pd_table(msg, pd);
+	if (ret)
+		goto out_free_msg;
+
+	genlmsg_end(msg, hdr);
+	return genlmsg_reply(msg, info);
+
+out_free_msg:
+	nlmsg_free(msg);
+	return ret;
 }
 
 static int __init em_netlink_init(void)
-- 
2.51.0

[PATCH RESEND v4 08/10] PM: EM: Implement em_notify_pd_deleted()

[Changwoo Min]

Add the event notification infrastructure and implement the event
notification for when a performance domain is deleted (EM_CMD_PD_DELETED).

The event contains the ID of the performance domain (EM_A_PD_TABLE_PD_ID)
so the userspace can identify the changed performance domain for further
processing.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/em_netlink.c | 56 +++++++++++++++++++++++++++++++++++++++
 kernel/power/em_netlink.h | 18 +++++++++++++
 2 files changed, 74 insertions(+)

diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
index 59953cfedf78..ff6aa848d998 100644
--- a/kernel/power/em_netlink.c
+++ b/kernel/power/em_netlink.c
@@ -213,6 +213,62 @@ int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
 	return ret;
 }
 
+
+/**************************** Event encoding *********************************/
+int em_notify_pd_created(const struct em_perf_domain *pd)
+{
+	return -EOPNOTSUPP;
+}
+
+int em_notify_pd_updated(const struct em_perf_domain *pd)
+{
+	return -EOPNOTSUPP;
+}
+
+static int __em_notify_pd_deleted_size(const struct em_perf_domain *pd)
+{
+	int id_sz = nla_total_size(sizeof(u32)); /* EM_A_PD_TABLE_PD_ID */
+
+	return nlmsg_total_size(genlmsg_msg_size(id_sz));
+}
+
+int em_notify_pd_deleted(const struct em_perf_domain *pd)
+{
+	struct sk_buff *msg;
+	int ret = -EMSGSIZE;
+	void *hdr;
+	int msg_sz;
+
+	if (!genl_has_listeners(&em_nl_family, &init_net, EM_NLGRP_EVENT))
+		return 0;
+
+	msg_sz = __em_notify_pd_deleted_size(pd);
+
+	msg = genlmsg_new(msg_sz, GFP_KERNEL);
+	if (!msg)
+		return -ENOMEM;
+
+	hdr = genlmsg_put(msg, 0, 0, &em_nl_family, 0, EM_CMD_PD_DELETED);
+	if (!hdr)
+		goto out_free_msg;
+
+	if (nla_put_u32(msg, EM_A_PD_TABLE_PD_ID, pd->id)) {
+		ret = -EMSGSIZE;
+		goto out_free_msg;
+	}
+
+	genlmsg_end(msg, hdr);
+
+	genlmsg_multicast(&em_nl_family, msg, 0, EM_NLGRP_EVENT, GFP_KERNEL);
+
+	return 0;
+
+out_free_msg:
+	nlmsg_free(msg);
+	return ret;
+}
+
+/**************************** Initialization *********************************/
 static int __init em_netlink_init(void)
 {
 	return genl_register_family(&em_nl_family);
diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
index acd186c92d6b..938c84ca1f40 100644
--- a/kernel/power/em_netlink.h
+++ b/kernel/power/em_netlink.h
@@ -10,7 +10,25 @@
 #define _EM_NETLINK_H
 
 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
+int em_notify_pd_created(const struct em_perf_domain *pd);
+int em_notify_pd_deleted(const struct em_perf_domain *pd);
+int em_notify_pd_updated(const struct em_perf_domain *pd);
+
 #else
+static inline int em_notify_pd_created(const struct em_perf_domain *pd)
+{
+	return 0;
+}
+
+static inline int em_notify_pd_deleted(const struct em_perf_domain *pd)
+{
+	return 0;
+}
+
+static inline int em_notify_pd_updated(const struct em_perf_domain *pd)
+{
+	return 0;
+}
 #endif
 
 #endif /* _EM_NETLINK_H */
-- 
2.51.0

[PATCH RESEND v4 09/10] PM: EM: Implement em_notify_pd_created/updated()

[Changwoo Min]

Implement two event notifications when a performance domain is created
(EM_CMD_PD_CREATED) and updated (EM_CMD_PD_UPDATED). The message format
of these two event notifications is the same as EM_CMD_GET_PD_TABLE --
containing the performance domain's ID and its energy model table.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/em_netlink.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
index ff6aa848d998..ff3eab078546 100644
--- a/kernel/power/em_netlink.c
+++ b/kernel/power/em_netlink.c
@@ -215,14 +215,48 @@ int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
 
 
 /**************************** Event encoding *********************************/
+static int __em_notify_pd_table(const struct em_perf_domain *pd, int ntf_type)
+{
+	struct sk_buff *msg;
+	int msg_sz, ret = -EMSGSIZE;
+	void *hdr;
+
+	if (!genl_has_listeners(&em_nl_family, &init_net, EM_NLGRP_EVENT))
+		return 0;
+
+	msg_sz = __em_nl_get_pd_table_size(pd);
+
+	msg = genlmsg_new(msg_sz, GFP_KERNEL);
+	if (!msg)
+		return -ENOMEM;
+
+	hdr = genlmsg_put(msg, 0, 0, &em_nl_family, 0, ntf_type);
+	if (!hdr)
+		goto out_free_msg;
+
+	ret = __em_nl_get_pd_table(msg, pd);
+	if (ret)
+		goto out_free_msg;
+
+	genlmsg_end(msg, hdr);
+
+	genlmsg_multicast(&em_nl_family, msg, 0, EM_NLGRP_EVENT, GFP_KERNEL);
+
+	return 0;
+
+out_free_msg:
+	nlmsg_free(msg);
+	return ret;
+}
+
 int em_notify_pd_created(const struct em_perf_domain *pd)
 {
-	return -EOPNOTSUPP;
+	return __em_notify_pd_table(pd, EM_CMD_PD_CREATED);
 }
 
 int em_notify_pd_updated(const struct em_perf_domain *pd)
 {
-	return -EOPNOTSUPP;
+	return __em_notify_pd_table(pd, EM_CMD_PD_UPDATED);
 }
 
 static int __em_notify_pd_deleted_size(const struct em_perf_domain *pd)
-- 
2.51.0

[PATCH RESEND v4 10/10] PM: EM: Notify an event when the performance domain changes

[Changwoo Min]

Send an event to userspace when a performance domain is created or deleted,
or its energy model is updated.

Signed-off-by: Changwoo Min <changwoo@igalia.com>
---
 kernel/power/energy_model.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
index 740076d24479..fca32d1c6661 100644
--- a/kernel/power/energy_model.c
+++ b/kernel/power/energy_model.c
@@ -17,6 +17,8 @@
 #include <linux/sched/topology.h>
 #include <linux/slab.h>
 
+#include "em_netlink.h"
+
 /*
  * Mutex serializing the registrations of performance domains and letting
  * callbacks defined by drivers sleep.
@@ -350,6 +352,8 @@ int em_dev_update_perf_domain(struct device *dev,
 	em_table_free(old_table);
 
 	mutex_unlock(&em_pd_mutex);
+
+	em_notify_pd_updated(pd);
 	return 0;
 }
 EXPORT_SYMBOL_GPL(em_dev_update_perf_domain);
@@ -697,6 +701,7 @@ int em_dev_register_pd_no_update(struct device *dev, unsigned int nr_states,
 	list_add_tail(&dev->em_pd->node, &em_pd_list);
 	mutex_unlock(&em_pd_list_mutex);
 
+	em_notify_pd_created(dev->em_pd);
 	return ret;
 }
 EXPORT_SYMBOL_GPL(em_dev_register_pd_no_update);
@@ -719,6 +724,8 @@ void em_dev_unregister_perf_domain(struct device *dev)
 	list_del_init(&dev->em_pd->node);
 	mutex_unlock(&em_pd_list_mutex);
 
+	em_notify_pd_deleted(dev->em_pd);
+
 	/*
 	 * The mutex separates all register/unregister requests and protects
 	 * from potential clean-up/setup issues in the debugfs directories.
-- 
2.51.0

Re: [PATCH RESEND v4 01/10] PM: EM: Assign a unique ID when creating a performance domain

[Lukasz Luba]

Hi Chanwoo,

My apologies to delay on this topic.

On 9/21/25 04:19, Changwoo Min wrote:
> It is necessary to refer to a specific performance domain from a
> userspace. For example, the energy model of a particular performance
> domain is updated.
> 
> To this end, assign a unique ID to each performance domain to address it,

Is this related to the sched_ext view on the EM that we cannot re-use
the allocated ID for the given domain?

> and manage them in a global linked list to look up a specific one by
> matching ID. IDA is used for ID assignment, and the mutex is used to
> protect the global list from concurrent access.
> 
> Note that the mutex (em_pd_list_mutex) is not supposed to hold while
> holding em_pd_mutex to avoid ABBA deadlock.

This might be tricky design, but I have seen in some other
patches you've added the lockdep, so we might have some safety net.

> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   include/linux/energy_model.h |  4 ++++
>   kernel/power/energy_model.c  | 33 ++++++++++++++++++++++++++++++++-
>   2 files changed, 36 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
> index 61d50571ad88..43aa6153dc57 100644
> --- a/include/linux/energy_model.h
> +++ b/include/linux/energy_model.h
> @@ -54,6 +54,8 @@ struct em_perf_table {
>   /**
>    * struct em_perf_domain - Performance domain
>    * @em_table:		Pointer to the runtime modifiable em_perf_table
> + * @node:		node in	em_pd_list (in energy_model.c)
> + * @id:			A unique ID number for each performance domain
>    * @nr_perf_states:	Number of performance states
>    * @min_perf_state:	Minimum allowed Performance State index
>    * @max_perf_state:	Maximum allowed Performance State index
> @@ -71,6 +73,8 @@ struct em_perf_table {
>    */
>   struct em_perf_domain {
>   	struct em_perf_table __rcu *em_table;
> +	struct list_head node;
> +	int id;
>   	int nr_perf_states;
>   	int min_perf_state;
>   	int max_perf_state;
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 8df55397414a..3fe562b6230e 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -23,6 +23,16 @@
>    */
>   static DEFINE_MUTEX(em_pd_mutex);
>   
> +/*
> + * Manage performance domains with IDs. One can iterate the performance domains
> + * through the list and pick one with their associated ID. The mutex serializes
> + * the list access. When holding em_pd_list_mutex, em_pd_mutex should not be
> + * taken to avoid potential deadlock.
> + */
> +static DEFINE_IDA(em_pd_ida);
> +static LIST_HEAD(em_pd_list);
> +static DEFINE_MUTEX(em_pd_list_mutex);
> +
>   static void em_cpufreq_update_efficiencies(struct device *dev,
>   					   struct em_perf_state *table);
>   static void em_check_capacity_update(void);
> @@ -396,7 +406,7 @@ static int em_create_pd(struct device *dev, int nr_states,
>   	struct em_perf_table *em_table;
>   	struct em_perf_domain *pd;
>   	struct device *cpu_dev;
> -	int cpu, ret, num_cpus;
> +	int cpu, ret, num_cpus, id;
>   
>   	if (_is_cpu_device(dev)) {
>   		num_cpus = cpumask_weight(cpus);
> @@ -420,6 +430,13 @@ static int em_create_pd(struct device *dev, int nr_states,
>   
>   	pd->nr_perf_states = nr_states;
>   
> +	INIT_LIST_HEAD(&pd->node);
> +
> +	id = ida_alloc(&em_pd_ida, GFP_KERNEL);
> +	if (id < 0)
> +		return -ENOMEM;
> +	pd->id = id;
> +
>   	em_table = em_table_alloc(pd);
>   	if (!em_table)
>   		goto free_pd;
> @@ -444,6 +461,7 @@ static int em_create_pd(struct device *dev, int nr_states,
>   	kfree(em_table);
>   free_pd:
>   	kfree(pd);
> +	ida_free(&em_pd_ida, id);
>   	return -EINVAL;
>   }
>   
> @@ -660,6 +678,13 @@ int em_dev_register_pd_no_update(struct device *dev, unsigned int nr_states,
>   unlock:
>   	mutex_unlock(&em_pd_mutex);
>   
> +	if (_is_cpu_device(dev))
> +		em_check_capacity_update();
> +
> +	mutex_lock(&em_pd_list_mutex);
> +	list_add_tail(&dev->em_pd->node, &em_pd_list);
> +	mutex_unlock(&em_pd_list_mutex);
> +
>   	return ret;
>   }
>   EXPORT_SYMBOL_GPL(em_dev_register_pd_no_update);
> @@ -678,6 +703,10 @@ void em_dev_unregister_perf_domain(struct device *dev)
>   	if (_is_cpu_device(dev))
>   		return;
>   
> +	mutex_lock(&em_pd_list_mutex);
> +	list_del_init(&dev->em_pd->node);
> +	mutex_unlock(&em_pd_list_mutex);
> +
>   	/*
>   	 * The mutex separates all register/unregister requests and protects
>   	 * from potential clean-up/setup issues in the debugfs directories.
> @@ -689,6 +718,8 @@ void em_dev_unregister_perf_domain(struct device *dev)
>   	em_table_free(rcu_dereference_protected(dev->em_pd->em_table,
>   						lockdep_is_held(&em_pd_mutex)));
>   
> +	ida_free(&em_pd_ida, dev->em_pd->id);
> +
>   	kfree(dev->em_pd);
>   	dev->em_pd = NULL;
>   	mutex_unlock(&em_pd_mutex);

Apart from that, the code itself looks sane.

Regards,
Lukasz

Re: [PATCH RESEND v4 01/10] PM: EM: Assign a unique ID when creating a performance domain

[Lukasz Luba]

On 10/6/25 09:17, Lukasz Luba wrote:
> Hi Chanwoo,
> 
> My apologies to delay on this topic.
> 
> On 9/21/25 04:19, Changwoo Min wrote:
>> It is necessary to refer to a specific performance domain from a
>> userspace. For example, the energy model of a particular performance
>> domain is updated.
>>
>> To this end, assign a unique ID to each performance domain to address it,
> 
> Is this related to the sched_ext view on the EM that we cannot re-use
> the allocated ID for the given domain?

Ignore that comment, I know the need now.

Although, there is a small code sneak below...



[..]

>> @@ -660,6 +678,13 @@ int em_dev_register_pd_no_update(struct device 
>> *dev, unsigned int nr_states,
>>   unlock:
>>       mutex_unlock(&em_pd_mutex);
>> +    if (_is_cpu_device(dev))
>> +        em_check_capacity_update();
>> +

It doesn't below to this $subject at all. It looks like
it was left from some your local changes, isn't it?

Re: [PATCH RESEND v4 02/10] PM: EM: Expose the ID of a performance domain via debugfs

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> For ease of debugging, let's expose the assigned ID of a performance
> domain through debugfs (e.g., /sys/kernel/debug/energy_model/cpu0/id).
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/energy_model.c | 12 ++++++++++++
>   1 file changed, 12 insertions(+)
> 
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 3fe562b6230e..8998a7f4910a 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -126,6 +126,16 @@ static int em_debug_flags_show(struct seq_file *s, void *unused)
>   }
>   DEFINE_SHOW_ATTRIBUTE(em_debug_flags);
>   
> +static int em_debug_id_show(struct seq_file *s, void *unused)
> +{
> +	struct em_perf_domain *pd = s->private;
> +
> +	seq_printf(s, "%d\n", pd->id);
> +
> +	return 0;
> +}
> +DEFINE_SHOW_ATTRIBUTE(em_debug_id);
> +
>   static void em_debug_create_pd(struct device *dev)
>   {
>   	struct em_dbg_info *em_dbg;
> @@ -142,6 +152,8 @@ static void em_debug_create_pd(struct device *dev)
>   	debugfs_create_file("flags", 0444, d, dev->em_pd,
>   			    &em_debug_flags_fops);
>   
> +	debugfs_create_file("id", 0444, d, dev->em_pd, &em_debug_id_fops);
> +
>   	em_dbg = devm_kcalloc(dev, dev->em_pd->nr_perf_states,
>   			      sizeof(*em_dbg), GFP_KERNEL);
>   	if (!em_dbg)

LGTM,

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH RESEND v4 03/10] PM: EM: Add an iterator and accessor for the performance domain

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Add an iterator function (for_each_em_perf_domain) that iterates all the
> performance domains in the global list. A passed callback function (cb) is
> called for each performance domain.
> 
> Additionally, add a lookup function (em_perf_domain_get_by_id) that
> searches for a performance domain by matching the ID in the global list.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   include/linux/energy_model.h | 15 +++++++++++++++
>   kernel/power/energy_model.c  | 34 ++++++++++++++++++++++++++++++++++
>   2 files changed, 49 insertions(+)
> 
> diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
> index 43aa6153dc57..21279e779188 100644
> --- a/include/linux/energy_model.h
> +++ b/include/linux/energy_model.h
> @@ -344,6 +344,10 @@ struct em_perf_state *em_perf_state_from_pd(struct em_perf_domain *pd)
>   	return rcu_dereference(pd->em_table)->state;
>   }
>   
> +int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
> +			    void *data);
> +struct em_perf_domain *em_perf_domain_get_by_id(int id);
> +

This doesn't have to go into this header.

>   #else
>   struct em_data_callback {};
>   #define EM_ADV_DATA_CB(_active_power_cb, _cost_cb) { }
> @@ -420,6 +424,17 @@ int em_update_performance_limits(struct em_perf_domain *pd,
>   }
>   static inline void em_adjust_cpu_capacity(unsigned int cpu) {}
>   static inline void em_rebuild_sched_domains(void) {}
> +static inline
> +int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
> +			    void *data)
> +{
> +	return -EINVAL;
> +}
> +static inline
> +struct em_perf_domain *em_perf_domain_get_by_id(int id)
> +{
> +	return NULL;
> +}
>   #endif


Please create a local helpers header:
kernel/power/em_helpers.h

and add these two declarations there. In that new
local header there is no need to implement the empty
inline functions as well (so would be simpler).

>   
>   #endif
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 8998a7f4910a..740076d24479 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -1000,3 +1000,37 @@ void em_rebuild_sched_domains(void)
>   	 */
>   	schedule_work(&rebuild_sd_work);
>   }
> +
> +int for_each_em_perf_domain(int (*cb)(struct em_perf_domain*, void *),
> +			    void *data)
> +{
> +	struct em_perf_domain *pd;
> +
> +	lockdep_assert_not_held(&em_pd_mutex);
> +	guard(mutex)(&em_pd_list_mutex);
> +
> +	list_for_each_entry(pd, &em_pd_list, node) {
> +		int ret;
> +
> +		ret = cb(pd, data);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +struct em_perf_domain *em_perf_domain_get_by_id(int id)
> +{
> +	struct em_perf_domain *pd;
> +
> +	lockdep_assert_not_held(&em_pd_mutex);
> +	guard(mutex)(&em_pd_list_mutex);
> +
> +	list_for_each_entry(pd, &em_pd_list, node) {
> +		if (pd->id == id)
> +			return pd;
> +	}
> +
> +	return NULL;
> +}

That code looks good, you can keep it. Although, please
add the comments above these functions that they are
only used locally as helpers for the notifications.

Re: [PATCH RESEND v4 04/10] PM: EM: Add em.yaml and autogen files

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Add a generic netlink spec in YAML format and autogenerate boilerplate
> code using ynl-regen.sh to introduce a generic netlink for the energy
> model. It allows a userspace program to read the performance domain and
> its energy model. It notifies the userspace program when a performance
> domain is created or deleted or its energy model is updated through a
> multicast interface.
> 
> Specifically, it supports two commands:
>    - EM_CMD_GET_PDS: Get the list of information for all performance
>      domains.
>    - EM_CMD_GET_PD_TABLE: Get the energy model table of a performance
>      domain.
> 
> Also, it supports three notification events:
>    - EM_CMD_PD_CREATED: When a performance domain is created.
>    - EM_CMD_PD_DELETED: When a performance domain is deleted.
>    - EM_CMD_PD_UPDATED: When the energy model table of a performance domain
>      is updated.
> 
> Finally, update MAINTAINERS to include new files.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   Documentation/netlink/specs/em.yaml | 113 ++++++++++++++++++++++++++++
>   MAINTAINERS                         |   3 +
>   include/uapi/linux/energy_model.h   |  62 +++++++++++++++
>   kernel/power/em_netlink_autogen.c   |  48 ++++++++++++
>   kernel/power/em_netlink_autogen.h   |  23 ++++++
>   5 files changed, 249 insertions(+)
>   create mode 100644 Documentation/netlink/specs/em.yaml
>   create mode 100644 include/uapi/linux/energy_model.h
>   create mode 100644 kernel/power/em_netlink_autogen.c
>   create mode 100644 kernel/power/em_netlink_autogen.h
> 
> diff --git a/Documentation/netlink/specs/em.yaml b/Documentation/netlink/specs/em.yaml
> new file mode 100644
> index 000000000000..9905ca482325
> --- /dev/null
> +++ b/Documentation/netlink/specs/em.yaml
> @@ -0,0 +1,113 @@
> +# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
> +
> +name: em
> +
> +doc: |
> +  Energy model netlink interface to notify its changes.
> +
> +protocol: genetlink
> +
> +uapi-header: linux/energy_model.h
> +
> +attribute-sets:
> +  -
> +    name: pds
> +    attributes:
> +      -
> +        name: pd
> +        type: nest
> +        nested-attributes: pd
> +        multi-attr: true
> +  -
> +    name: pd
> +    attributes:
> +      -
> +        name: pad
> +        type: pad
> +      -
> +        name: pd-id
> +        type: u32
> +      -
> +        name: flags
> +        type: u64
> +      -
> +        name: cpus
> +        type: string
> +  -
> +    name: pd-table
> +    attributes:
> +      -
> +        name: pd-id
> +        type: u32
> +      -
> +        name: ps
> +        type: nest
> +        nested-attributes: ps
> +        multi-attr: true
> +  -
> +    name: ps
> +    attributes:
> +      -
> +        name: pad
> +        type: pad
> +      -
> +        name: performance
> +        type: u64
> +      -
> +        name: frequency
> +        type: u64
> +      -
> +        name: power
> +        type: u64
> +      -
> +        name: cost
> +        type: u64
> +      -
> +        name: flags
> +        type: u64
> +
> +operations:
> +  list:
> +    -
> +      name: get-pds
> +      attribute-set: pds
> +      doc: Get the list of information for all performance domains.
> +      do:
> +        reply:
> +          attributes:
> +            - pd
> +    -
> +      name: get-pd-table
> +      attribute-set: pd-table
> +      doc: Get the energy model table of a performance domain.
> +      do:
> +        request:
> +          attributes:
> +            - pd-id
> +        reply:
> +          attributes:
> +            - pd-id
> +            - ps
> +    -
> +      name: pd-created
> +      doc: A performance domain is created.
> +      notify: get-pd-table
> +      mcgrp: event
> +    -
> +      name: pd-updated
> +      doc: A performance domain is updated.
> +      notify: get-pd-table
> +      mcgrp: event
> +    -
> +      name: pd-deleted
> +      doc: A performance domain is deleted.
> +      attribute-set: pd-table
> +      event:
> +        attributes:
> +            - pd-id
> +      mcgrp: event
> +
> +mcast-groups:
> +  list:
> +    -
> +      name: event
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 520fb4e379a3..0992029d271d 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -9032,6 +9032,9 @@ S:	Maintained
>   F:	kernel/power/energy_model.c
>   F:	include/linux/energy_model.h
>   F:	Documentation/power/energy-model.rst
> +F:	Documentation/netlink/specs/em.yaml
> +F:	include/uapi/linux/energy_model.h
> +F:	kernel/power/em_netlink_autogen.*
>   
>   EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
>   M:	Laurentiu Tudor <laurentiu.tudor@nxp.com>
> diff --git a/include/uapi/linux/energy_model.h b/include/uapi/linux/energy_model.h
> new file mode 100644
> index 000000000000..4ec4c0eabbbb
> --- /dev/null
> +++ b/include/uapi/linux/energy_model.h
> @@ -0,0 +1,62 @@
> +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
> +/* Do not edit directly, auto-generated from: */
> +/*	Documentation/netlink/specs/em.yaml */
> +/* YNL-GEN uapi header */
> +
> +#ifndef _UAPI_LINUX_ENERGY_MODEL_H
> +#define _UAPI_LINUX_ENERGY_MODEL_H
> +
> +#define EM_FAMILY_NAME		"em"
> +#define EM_FAMILY_VERSION	1
> +
> +enum {
> +	EM_A_PDS_PD = 1,
> +
> +	__EM_A_PDS_MAX,
> +	EM_A_PDS_MAX = (__EM_A_PDS_MAX - 1)
> +};
> +
> +enum {
> +	EM_A_PD_PAD = 1,
> +	EM_A_PD_PD_ID,
> +	EM_A_PD_FLAGS,
> +	EM_A_PD_CPUS,
> +
> +	__EM_A_PD_MAX,
> +	EM_A_PD_MAX = (__EM_A_PD_MAX - 1)
> +};
> +
> +enum {
> +	EM_A_PD_TABLE_PD_ID = 1,
> +	EM_A_PD_TABLE_PS,
> +
> +	__EM_A_PD_TABLE_MAX,
> +	EM_A_PD_TABLE_MAX = (__EM_A_PD_TABLE_MAX - 1)
> +};
> +
> +enum {
> +	EM_A_PS_PAD = 1,
> +	EM_A_PS_PERFORMANCE,
> +	EM_A_PS_FREQUENCY,
> +	EM_A_PS_POWER,
> +	EM_A_PS_COST,
> +	EM_A_PS_FLAGS,
> +
> +	__EM_A_PS_MAX,
> +	EM_A_PS_MAX = (__EM_A_PS_MAX - 1)
> +};
> +
> +enum {
> +	EM_CMD_GET_PDS = 1,
> +	EM_CMD_GET_PD_TABLE,
> +	EM_CMD_PD_CREATED,
> +	EM_CMD_PD_UPDATED,
> +	EM_CMD_PD_DELETED,
> +
> +	__EM_CMD_MAX,
> +	EM_CMD_MAX = (__EM_CMD_MAX - 1)
> +};
> +
> +#define EM_MCGRP_EVENT	"event"
> +
> +#endif /* _UAPI_LINUX_ENERGY_MODEL_H */
> diff --git a/kernel/power/em_netlink_autogen.c b/kernel/power/em_netlink_autogen.c
> new file mode 100644
> index 000000000000..a7a09ab1d1c2
> --- /dev/null
> +++ b/kernel/power/em_netlink_autogen.c
> @@ -0,0 +1,48 @@
> +// SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
> +/* Do not edit directly, auto-generated from: */
> +/*	Documentation/netlink/specs/em.yaml */
> +/* YNL-GEN kernel source */
> +
> +#include <net/netlink.h>
> +#include <net/genetlink.h>
> +
> +#include "em_netlink_autogen.h"
> +
> +#include <uapi/linux/energy_model.h>
> +
> +/* EM_CMD_GET_PD_TABLE - do */
> +static const struct nla_policy em_get_pd_table_nl_policy[EM_A_PD_TABLE_PD_ID + 1] = {
> +	[EM_A_PD_TABLE_PD_ID] = { .type = NLA_U32, },
> +};
> +
> +/* Ops table for em */
> +static const struct genl_split_ops em_nl_ops[] = {
> +	{
> +		.cmd	= EM_CMD_GET_PDS,
> +		.doit	= em_nl_get_pds_doit,
> +		.flags	= GENL_CMD_CAP_DO,
> +	},
> +	{
> +		.cmd		= EM_CMD_GET_PD_TABLE,
> +		.doit		= em_nl_get_pd_table_doit,
> +		.policy		= em_get_pd_table_nl_policy,
> +		.maxattr	= EM_A_PD_TABLE_PD_ID,
> +		.flags		= GENL_CMD_CAP_DO,
> +	},
> +};
> +
> +static const struct genl_multicast_group em_nl_mcgrps[] = {
> +	[EM_NLGRP_EVENT] = { "event", },
> +};
> +
> +struct genl_family em_nl_family __ro_after_init = {
> +	.name		= EM_FAMILY_NAME,
> +	.version	= EM_FAMILY_VERSION,
> +	.netnsok	= true,
> +	.parallel_ops	= true,
> +	.module		= THIS_MODULE,
> +	.split_ops	= em_nl_ops,
> +	.n_split_ops	= ARRAY_SIZE(em_nl_ops),
> +	.mcgrps		= em_nl_mcgrps,
> +	.n_mcgrps	= ARRAY_SIZE(em_nl_mcgrps),
> +};
> diff --git a/kernel/power/em_netlink_autogen.h b/kernel/power/em_netlink_autogen.h
> new file mode 100644
> index 000000000000..78ce609641f1
> --- /dev/null
> +++ b/kernel/power/em_netlink_autogen.h
> @@ -0,0 +1,23 @@
> +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
> +/* Do not edit directly, auto-generated from: */
> +/*	Documentation/netlink/specs/em.yaml */
> +/* YNL-GEN kernel header */
> +
> +#ifndef _LINUX_EM_GEN_H
> +#define _LINUX_EM_GEN_H
> +
> +#include <net/netlink.h>
> +#include <net/genetlink.h>
> +
> +#include <uapi/linux/energy_model.h>
> +
> +int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info);
> +int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info);
> +
> +enum {
> +	EM_NLGRP_EVENT,
> +};
> +
> +extern struct genl_family em_nl_family;
> +
> +#endif /* _LINUX_EM_GEN_H */


LGTM,

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH RESEND v4 05/10] PM: EM: Add a skeleton code for netlink notification

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Add a boilerplate code for netlink notification to register the new
> protocol family. Also, initialize and register the netlink during booting.
> The initialization is called at the postcore level, which is late enough
> after the generic netlink is initialized.
> 
> Finally, update MAINTAINERS to include new files.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   MAINTAINERS               |  2 +-
>   kernel/power/Makefile     |  5 ++++-
>   kernel/power/em_netlink.c | 35 +++++++++++++++++++++++++++++++++++
>   kernel/power/em_netlink.h | 16 ++++++++++++++++
>   4 files changed, 56 insertions(+), 2 deletions(-)
>   create mode 100644 kernel/power/em_netlink.c
>   create mode 100644 kernel/power/em_netlink.h
> 
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 0992029d271d..ba528836eac1 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -9034,7 +9034,7 @@ F:	include/linux/energy_model.h
>   F:	Documentation/power/energy-model.rst
>   F:	Documentation/netlink/specs/em.yaml
>   F:	include/uapi/linux/energy_model.h
> -F:	kernel/power/em_netlink_autogen.*
> +F:	kernel/power/em_netlink*.*
>   
>   EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
>   M:	Laurentiu Tudor <laurentiu.tudor@nxp.com>
> diff --git a/kernel/power/Makefile b/kernel/power/Makefile
> index 874ad834dc8d..284a760aade7 100644
> --- a/kernel/power/Makefile
> +++ b/kernel/power/Makefile
> @@ -21,4 +21,7 @@ obj-$(CONFIG_PM_WAKELOCKS)	+= wakelock.o
>   
>   obj-$(CONFIG_MAGIC_SYSRQ)	+= poweroff.o
>   
> -obj-$(CONFIG_ENERGY_MODEL)	+= energy_model.o
> +obj-$(CONFIG_ENERGY_MODEL)	+= em.o
> +em-y				:= energy_model.o
> +em-$(CONFIG_NET)		+= em_netlink_autogen.o em_netlink.o
> +
> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
> new file mode 100644
> index 000000000000..f3fbfeff29a4
> --- /dev/null
> +++ b/kernel/power/em_netlink.c
> @@ -0,0 +1,35 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + *
> + * Generic netlink for energy model.
> + *
> + * Copyright (c) 2025 Valve Corporation.
> + * Author: Changwoo Min <changwoo@igalia.com>
> + */
> +
> +#define pr_fmt(fmt) "energy_model: " fmt
> +
> +#include <linux/energy_model.h>
> +#include <net/sock.h>
> +#include <net/genetlink.h>
> +#include <uapi/linux/energy_model.h>
> +
> +#include "em_netlink.h"
> +#include "em_netlink_autogen.h"
> +
> +int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
> +{
> +	return -EOPNOTSUPP;
> +}
> +
> +int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
> +{
> +	return -EOPNOTSUPP;
> +}
> +
> +static int __init em_netlink_init(void)
> +{
> +	return genl_register_family(&em_nl_family);
> +}
> +postcore_initcall(em_netlink_init);
> +
> diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
> new file mode 100644
> index 000000000000..acd186c92d6b
> --- /dev/null
> +++ b/kernel/power/em_netlink.h
> @@ -0,0 +1,16 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + *
> + * Generic netlink for energy model.
> + *
> + * Copyright (c) 2025 Valve Corporation.
> + * Author: Changwoo Min <changwoo@igalia.com>
> + */
> +#ifndef _EM_NETLINK_H
> +#define _EM_NETLINK_H
> +
> +#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
> +#else
> +#endif
> +
> +#endif /* _EM_NETLINK_H */

Actually, those declarations of functions from patch 3/10 can
live in this header. We would avoid creating more local headers
in such case.

Then the patch 3/10 would have to go after this patch when
this header is introduced.

Please ignore the comment in the patch 3/10 and try to
use this header. It is also logically linked to the
notifications, so belongs to such header IMHO.

Re: [PATCH RESEND v4 06/10] PM: EM: Implement em_nl_get_pds_doit()

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> When a userspace requests EM_CMD_GET_PDS, the kernel responds with
> information on all performance domains. The message format of the
> response is as follows:
> 
> EM_A_PDS_PD (NLA_NESTED)*
>      EM_A_PD_PD_ID (NLA_U32)
>      EM_A_PD_FLAGS (NLA_U64)
>      EM_A_PD_CPUS (NLA_STRING)
> 
> where EM_A_PDS_PD can be repeated as many times as there are performance
> domains, and EM_A_PD_CPUS is a hexadecimal string representing a CPU
> bitmask.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/em_netlink.c | 82 ++++++++++++++++++++++++++++++++++++++-
>   1 file changed, 81 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
> index f3fbfeff29a4..31b27c6fe3c9 100644
> --- a/kernel/power/em_netlink.c
> +++ b/kernel/power/em_netlink.c
> @@ -17,9 +17,89 @@
>   #include "em_netlink.h"
>   #include "em_netlink_autogen.h"
>   
> +#define EM_A_PD_CPUS_LEN		256
> +
> +/*************************** Command encoding ********************************/
> +static int __em_nl_get_pd_size(struct em_perf_domain *pd, void *data)
> +{
> +	char cpus_buf[EM_A_PD_CPUS_LEN];
> +	int *tot_msg_sz = data;
> +	int msg_sz, cpus_sz;
> +
> +	cpus_sz = snprintf(cpus_buf, sizeof(cpus_buf), "%*pb",
> +			   cpumask_pr_args(to_cpumask(pd->cpus)));
> +
> +	msg_sz = nla_total_size(0) +			/* EM_A_PDS_PD */
> +		 nla_total_size(sizeof(u32)) +		/* EM_A_PD_PD_ID */
> +		 nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PD_FLAGS */
> +		 nla_total_size(cpus_sz);		/* EM_A_PD_CPUS */
> +
> +	*tot_msg_sz += nlmsg_total_size(genlmsg_msg_size(msg_sz));
> +	return 0;
> +}
> +
> +static int __em_nl_get_pd(struct em_perf_domain *pd, void *data)
> +{
> +	char cpus_buf[EM_A_PD_CPUS_LEN];
> +	struct sk_buff *msg = data;
> +	struct nlattr *entry;
> +
> +	entry = nla_nest_start(msg, EM_A_PDS_PD);
> +	if (!entry)
> +		goto out_cancel_nest;
> +
> +	if (nla_put_u32(msg, EM_A_PD_PD_ID, pd->id))
> +		goto out_cancel_nest;
> +
> +	if (nla_put_u64_64bit(msg, EM_A_PD_FLAGS, pd->flags, EM_A_PD_PAD))
> +		goto out_cancel_nest;
> +
> +	snprintf(cpus_buf, sizeof(cpus_buf), "%*pb",
> +		 cpumask_pr_args(to_cpumask(pd->cpus)));
> +	if (nla_put_string(msg, EM_A_PD_CPUS, cpus_buf))
> +		goto out_cancel_nest;
> +
> +	nla_nest_end(msg, entry);
> +
> +	return 0;
> +
> +out_cancel_nest:
> +	nla_nest_cancel(msg, entry);
> +
> +	return -EMSGSIZE;
> +}
> +
>   int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
>   {
> -	return -EOPNOTSUPP;
> +	struct sk_buff *msg;
> +	void *hdr;
> +	int cmd = info->genlhdr->cmd;
> +	int ret = -EMSGSIZE, msg_sz = 0;
> +
> +	for_each_em_perf_domain(__em_nl_get_pd_size, &msg_sz);
> +
> +	msg = genlmsg_new(msg_sz, GFP_KERNEL);
> +	if (!msg)
> +		return -ENOMEM;
> +
> +	hdr = genlmsg_put_reply(msg, info, &em_nl_family, 0, cmd);
> +	if (!hdr)
> +		goto out_free_msg;
> +
> +	ret = for_each_em_perf_domain(__em_nl_get_pd, msg);
> +	if (ret)
> +		goto out_cancel_msg;
> +
> +	genlmsg_end(msg, hdr);
> +
> +	return genlmsg_reply(msg, info);
> +
> +out_cancel_msg:
> +	genlmsg_cancel(msg, hdr);
> +out_free_msg:
> +	nlmsg_free(msg);
> +
> +	return ret;
>   }
>   
>   int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)

LGTM,

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH RESEND v4 07/10] PM: EM: Implement em_nl_get_pd_table_doit()

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> When a userspace requests EM_CMD_GET_PD_TABLE with an ID of a performancei

s/performancei/performance/

> domain, the kernel reports back the energy model table of the specified
> performance domain. The message format of the response is as follows:
> 
> EM_A_PD_TABLE_PD_ID (NLA_U32)
> EM_A_PD_TABLE_PS (NLA_NESTED)*
>      EM_A_PS_PERFORMANCE (NLA_U64)
>      EM_A_PS_FREQUENCY (NLA_U64)
>      EM_A_PS_POWER (NLA_U64)
>      EM_A_PS_COST (NLA_U64)
>      EM_A_PS_FLAGS (NLA_U64)
> 
> where EM_A_PD_TABLE_PS can be repeated as many times as there are
> performance states (struct em_perf_state).
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/em_netlink.c | 108 +++++++++++++++++++++++++++++++++++++-
>   1 file changed, 107 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
> index 31b27c6fe3c9..59953cfedf78 100644
> --- a/kernel/power/em_netlink.c
> +++ b/kernel/power/em_netlink.c
> @@ -102,9 +102,115 @@ int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
>   	return ret;
>   }
>   
> +static struct em_perf_domain *__em_nl_get_pd_table_id(struct nlattr **attrs)
> +{
> +	struct em_perf_domain *pd;
> +	int id;
> +
> +	if (!attrs[EM_A_PD_TABLE_PD_ID])
> +		return NULL;
> +
> +	id = nla_get_u32(attrs[EM_A_PD_TABLE_PD_ID]);
> +	pd = em_perf_domain_get_by_id(id);
> +	return pd;
> +}
> +
> +static int __em_nl_get_pd_table_size(const struct em_perf_domain *pd)
> +{
> +	int id_sz, ps_sz;
> +
> +	id_sz = nla_total_size(sizeof(u32));		/* EM_A_PD_TABLE_PD_ID */
> +	ps_sz = nla_total_size(0) +			/* EM_A_PD_TABLE_PS */
> +		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_PERFORMANCE */
> +		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_FREQUENCY */
> +		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_POWER */
> +		nla_total_size_64bit(sizeof(u64)) +	/* EM_A_PS_COST */
> +		nla_total_size_64bit(sizeof(u64));	/* EM_A_PS_FLAGS */
> +	ps_sz *= pd->nr_perf_states;
> +
> +	return nlmsg_total_size(genlmsg_msg_size(id_sz + ps_sz));
> +}
> +
> +static int __em_nl_get_pd_table(struct sk_buff *msg, const struct em_perf_domain *pd)
> +{
> +	struct em_perf_state *table, *ps;
> +	struct nlattr *entry;
> +	int i;
> +
> +	if (nla_put_u32(msg, EM_A_PD_TABLE_PD_ID, pd->id))
> +		goto out_err;
> +
> +	rcu_read_lock();
> +	table = em_perf_state_from_pd((struct em_perf_domain *)pd);
> +
> +	for (i = 0; i < pd->nr_perf_states; i++) {
> +		ps = &table[i];
> +
> +		entry = nla_nest_start(msg, EM_A_PD_TABLE_PS);
> +		if (!entry)
> +			goto out_unlock_ps;
> +
> +		if (nla_put_u64_64bit(msg, EM_A_PS_PERFORMANCE,
> +				      ps->performance, EM_A_PS_PAD))
> +			goto out_cancel_ps_nest;
> +		if (nla_put_u64_64bit(msg, EM_A_PS_FREQUENCY,
> +				      ps->frequency, EM_A_PS_PAD))
> +			goto out_cancel_ps_nest;
> +		if (nla_put_u64_64bit(msg, EM_A_PS_POWER,
> +				      ps->power, EM_A_PS_PAD))
> +			goto out_cancel_ps_nest;
> +		if (nla_put_u64_64bit(msg, EM_A_PS_COST,
> +				      ps->cost, EM_A_PS_PAD))
> +			goto out_cancel_ps_nest;
> +		if (nla_put_u64_64bit(msg, EM_A_PS_FLAGS,
> +				      ps->flags, EM_A_PS_PAD))
> +			goto out_cancel_ps_nest;
> +
> +		nla_nest_end(msg, entry);
> +	}
> +	rcu_read_unlock();
> +	return 0;
> +
> +out_cancel_ps_nest:
> +	nla_nest_cancel(msg, entry);
> +out_unlock_ps:
> +	rcu_read_unlock();
> +out_err:
> +	return -EMSGSIZE;
> +}
> +
>   int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
>   {
> -	return -EOPNOTSUPP;
> +	struct sk_buff *msg;
> +	struct em_perf_domain *pd;
> +	void *hdr;
> +	int cmd = info->genlhdr->cmd;
> +	int msg_sz, ret = -EMSGSIZE;

Please put them in the reverse christmas tree order.

> +
> +	pd = __em_nl_get_pd_table_id(info->attrs);
> +	if (!pd)
> +		return -EINVAL;
> +
> +	msg_sz = __em_nl_get_pd_table_size(pd);
> +
> +	msg = genlmsg_new(msg_sz, GFP_KERNEL);
> +	if (!msg)
> +		return -ENOMEM;
> +
> +	hdr = genlmsg_put_reply(msg, info, &em_nl_family, 0, cmd);
> +	if (!hdr)
> +		goto out_free_msg;
> +
> +	ret = __em_nl_get_pd_table(msg, pd);
> +	if (ret)
> +		goto out_free_msg;
> +
> +	genlmsg_end(msg, hdr);
> +	return genlmsg_reply(msg, info);
> +
> +out_free_msg:
> +	nlmsg_free(msg);
> +	return ret;
>   }
>   
>   static int __init em_netlink_init(void)

After fixing that cosmetic thing you can add:

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH RESEND v4 08/10] PM: EM: Implement em_notify_pd_deleted()

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Add the event notification infrastructure and implement the event
> notification for when a performance domain is deleted (EM_CMD_PD_DELETED).
> 
> The event contains the ID of the performance domain (EM_A_PD_TABLE_PD_ID)
> so the userspace can identify the changed performance domain for further
> processing.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/em_netlink.c | 56 +++++++++++++++++++++++++++++++++++++++
>   kernel/power/em_netlink.h | 18 +++++++++++++
>   2 files changed, 74 insertions(+)
> 
> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
> index 59953cfedf78..ff6aa848d998 100644
> --- a/kernel/power/em_netlink.c
> +++ b/kernel/power/em_netlink.c
> @@ -213,6 +213,62 @@ int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
>   	return ret;
>   }
>   
> +
> +/**************************** Event encoding *********************************/
> +int em_notify_pd_created(const struct em_perf_domain *pd)
> +{
> +	return -EOPNOTSUPP;
> +}
> +
> +int em_notify_pd_updated(const struct em_perf_domain *pd)
> +{
> +	return -EOPNOTSUPP;
> +}

Those two functions doesn't belong to this patch $subject.
I would recommend adding them in the following patch where they
are properly defined.

> +
> +static int __em_notify_pd_deleted_size(const struct em_perf_domain *pd)
> +{
> +	int id_sz = nla_total_size(sizeof(u32)); /* EM_A_PD_TABLE_PD_ID */
> +
> +	return nlmsg_total_size(genlmsg_msg_size(id_sz));
> +}
> +
> +int em_notify_pd_deleted(const struct em_perf_domain *pd)
> +{
> +	struct sk_buff *msg;
> +	int ret = -EMSGSIZE;
> +	void *hdr;
> +	int msg_sz;
> +
> +	if (!genl_has_listeners(&em_nl_family, &init_net, EM_NLGRP_EVENT))
> +		return 0;
> +
> +	msg_sz = __em_notify_pd_deleted_size(pd);
> +
> +	msg = genlmsg_new(msg_sz, GFP_KERNEL);
> +	if (!msg)
> +		return -ENOMEM;
> +
> +	hdr = genlmsg_put(msg, 0, 0, &em_nl_family, 0, EM_CMD_PD_DELETED);
> +	if (!hdr)
> +		goto out_free_msg;
> +
> +	if (nla_put_u32(msg, EM_A_PD_TABLE_PD_ID, pd->id)) {
> +		ret = -EMSGSIZE;
> +		goto out_free_msg;
> +	}
> +
> +	genlmsg_end(msg, hdr);
> +
> +	genlmsg_multicast(&em_nl_family, msg, 0, EM_NLGRP_EVENT, GFP_KERNEL);
> +
> +	return 0;
> +
> +out_free_msg:
> +	nlmsg_free(msg);
> +	return ret;
> +}

It looks like we don't handle the error outputs of those
em_notify_* function (based on last patch in the set.

Can we just simply change the function to void?
It would also avoid some spurious static analysis tools checking the
code in future.

Re: [PATCH RESEND v4 09/10] PM: EM: Implement em_notify_pd_created/updated()

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Implement two event notifications when a performance domain is created
> (EM_CMD_PD_CREATED) and updated (EM_CMD_PD_UPDATED). The message format
> of these two event notifications is the same as EM_CMD_GET_PD_TABLE --
> containing the performance domain's ID and its energy model table.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/em_netlink.c | 38 ++++++++++++++++++++++++++++++++++++--
>   1 file changed, 36 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
> index ff6aa848d998..ff3eab078546 100644
> --- a/kernel/power/em_netlink.c
> +++ b/kernel/power/em_netlink.c
> @@ -215,14 +215,48 @@ int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
>   
>   
>   /**************************** Event encoding *********************************/
> +static int __em_notify_pd_table(const struct em_perf_domain *pd, int ntf_type)
> +{
> +	struct sk_buff *msg;
> +	int msg_sz, ret = -EMSGSIZE;
> +	void *hdr;
> +
> +	if (!genl_has_listeners(&em_nl_family, &init_net, EM_NLGRP_EVENT))
> +		return 0;
> +
> +	msg_sz = __em_nl_get_pd_table_size(pd);
> +
> +	msg = genlmsg_new(msg_sz, GFP_KERNEL);
> +	if (!msg)
> +		return -ENOMEM;
> +
> +	hdr = genlmsg_put(msg, 0, 0, &em_nl_family, 0, ntf_type);
> +	if (!hdr)
> +		goto out_free_msg;
> +
> +	ret = __em_nl_get_pd_table(msg, pd);
> +	if (ret)
> +		goto out_free_msg;
> +
> +	genlmsg_end(msg, hdr);
> +
> +	genlmsg_multicast(&em_nl_family, msg, 0, EM_NLGRP_EVENT, GFP_KERNEL);
> +
> +	return 0;
> +
> +out_free_msg:
> +	nlmsg_free(msg);
> +	return ret;
> +}
> +
>   int em_notify_pd_created(const struct em_perf_domain *pd)
>   {
> -	return -EOPNOTSUPP;
> +	return __em_notify_pd_table(pd, EM_CMD_PD_CREATED);
>   }
>   
>   int em_notify_pd_updated(const struct em_perf_domain *pd)
>   {
> -	return -EOPNOTSUPP;
> +	return __em_notify_pd_table(pd, EM_CMD_PD_UPDATED);
>   }
>   
>   static int __em_notify_pd_deleted_size(const struct em_perf_domain *pd)

Similar comment to the one in previous patch.

Can we just simply change these functions to void?

Re: [PATCH RESEND v4 10/10] PM: EM: Notify an event when the performance domain changes

[Lukasz Luba]

On 9/21/25 04:19, Changwoo Min wrote:
> Send an event to userspace when a performance domain is created or deleted,
> or its energy model is updated.
> 
> Signed-off-by: Changwoo Min <changwoo@igalia.com>
> ---
>   kernel/power/energy_model.c | 7 +++++++
>   1 file changed, 7 insertions(+)
> 
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 740076d24479..fca32d1c6661 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -17,6 +17,8 @@
>   #include <linux/sched/topology.h>
>   #include <linux/slab.h>
>   
> +#include "em_netlink.h"
> +
>   /*
>    * Mutex serializing the registrations of performance domains and letting
>    * callbacks defined by drivers sleep.
> @@ -350,6 +352,8 @@ int em_dev_update_perf_domain(struct device *dev,
>   	em_table_free(old_table);
>   
>   	mutex_unlock(&em_pd_mutex);
> +
> +	em_notify_pd_updated(pd);
>   	return 0;
>   }
>   EXPORT_SYMBOL_GPL(em_dev_update_perf_domain);
> @@ -697,6 +701,7 @@ int em_dev_register_pd_no_update(struct device *dev, unsigned int nr_states,
>   	list_add_tail(&dev->em_pd->node, &em_pd_list);
>   	mutex_unlock(&em_pd_list_mutex);
>   
> +	em_notify_pd_created(dev->em_pd);
>   	return ret;
>   }
>   EXPORT_SYMBOL_GPL(em_dev_register_pd_no_update);
> @@ -719,6 +724,8 @@ void em_dev_unregister_perf_domain(struct device *dev)
>   	list_del_init(&dev->em_pd->node);
>   	mutex_unlock(&em_pd_list_mutex);
>   
> +	em_notify_pd_deleted(dev->em_pd);
> +
>   	/*
>   	 * The mutex separates all register/unregister requests and protects
>   	 * from potential clean-up/setup issues in the debugfs directories.

LGTM,

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH RESEND v4 01/10] PM: EM: Assign a unique ID when creating a performance domain

[Changwoo Min]

Hi Lukasz,

On 10/6/25 21:24, Lukasz Luba wrote:
> 
> 
> On 10/6/25 09:17, Lukasz Luba wrote:
>> Hi Chanwoo,
>>
>> My apologies to delay on this topic.

Thank you for finding time and making an effort for the review! I
understand that it is not always possible to make time for review. :-)

>>
>> On 9/21/25 04:19, Changwoo Min wrote:
>>> It is necessary to refer to a specific performance domain from a
>>> userspace. For example, the energy model of a particular performance
>>> domain is updated.
>>>
>>> To this end, assign a unique ID to each performance domain to address 
>>> it,
>>
>> Is this related to the sched_ext view on the EM that we cannot re-use
>> the allocated ID for the given domain?
> 
> Ignore that comment, I know the need now.
> 
> Although, there is a small code sneak below...
> 
> 
> 
> [..]
> 
>>> @@ -660,6 +678,13 @@ int em_dev_register_pd_no_update(struct device 
>>> *dev, unsigned int nr_states,
>>>   unlock:
>>>       mutex_unlock(&em_pd_mutex);
>>> +    if (_is_cpu_device(dev))
>>> +        em_check_capacity_update();
>>> +
> 
> It doesn't below to this $subject at all. It looks like
> it was left from some your local changes, isn't it?

You are right. The code is redundant since the same check is done at
em_dev_register_perf_domain(). It is the side-effect of a bad rebase. I
will remove this in the next version.

Regards,
Changwoo Min

Re: [PATCH RESEND v4 05/10] PM: EM: Add a skeleton code for netlink notification

[Changwoo Min]

On 10/7/25 00:44, Lukasz Luba wrote:
> 
> 
> On 9/21/25 04:19, Changwoo Min wrote:
>> Add a boilerplate code for netlink notification to register the new
>> protocol family. Also, initialize and register the netlink during 
>> booting.
>> The initialization is called at the postcore level, which is late enough
>> after the generic netlink is initialized.
>>
>> Finally, update MAINTAINERS to include new files.
>>
>> Signed-off-by: Changwoo Min <changwoo@igalia.com>
>> ---
>>   MAINTAINERS               |  2 +-
>>   kernel/power/Makefile     |  5 ++++-
>>   kernel/power/em_netlink.c | 35 +++++++++++++++++++++++++++++++++++
>>   kernel/power/em_netlink.h | 16 ++++++++++++++++
>>   4 files changed, 56 insertions(+), 2 deletions(-)
>>   create mode 100644 kernel/power/em_netlink.c
>>   create mode 100644 kernel/power/em_netlink.h
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 0992029d271d..ba528836eac1 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -9034,7 +9034,7 @@ F:    include/linux/energy_model.h
>>   F:    Documentation/power/energy-model.rst
>>   F:    Documentation/netlink/specs/em.yaml
>>   F:    include/uapi/linux/energy_model.h
>> -F:    kernel/power/em_netlink_autogen.*
>> +F:    kernel/power/em_netlink*.*
>>   EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
>>   M:    Laurentiu Tudor <laurentiu.tudor@nxp.com>
>> diff --git a/kernel/power/Makefile b/kernel/power/Makefile
>> index 874ad834dc8d..284a760aade7 100644
>> --- a/kernel/power/Makefile
>> +++ b/kernel/power/Makefile
>> @@ -21,4 +21,7 @@ obj-$(CONFIG_PM_WAKELOCKS)    += wakelock.o
>>   obj-$(CONFIG_MAGIC_SYSRQ)    += poweroff.o
>> -obj-$(CONFIG_ENERGY_MODEL)    += energy_model.o
>> +obj-$(CONFIG_ENERGY_MODEL)    += em.o
>> +em-y                := energy_model.o
>> +em-$(CONFIG_NET)        += em_netlink_autogen.o em_netlink.o
>> +
>> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
>> new file mode 100644
>> index 000000000000..f3fbfeff29a4
>> --- /dev/null
>> +++ b/kernel/power/em_netlink.c
>> @@ -0,0 +1,35 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + *
>> + * Generic netlink for energy model.
>> + *
>> + * Copyright (c) 2025 Valve Corporation.
>> + * Author: Changwoo Min <changwoo@igalia.com>
>> + */
>> +
>> +#define pr_fmt(fmt) "energy_model: " fmt
>> +
>> +#include <linux/energy_model.h>
>> +#include <net/sock.h>
>> +#include <net/genetlink.h>
>> +#include <uapi/linux/energy_model.h>
>> +
>> +#include "em_netlink.h"
>> +#include "em_netlink_autogen.h"
>> +
>> +int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
>> +{
>> +    return -EOPNOTSUPP;
>> +}
>> +
>> +int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info *info)
>> +{
>> +    return -EOPNOTSUPP;
>> +}
>> +
>> +static int __init em_netlink_init(void)
>> +{
>> +    return genl_register_family(&em_nl_family);
>> +}
>> +postcore_initcall(em_netlink_init);
>> +
>> diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
>> new file mode 100644
>> index 000000000000..acd186c92d6b
>> --- /dev/null
>> +++ b/kernel/power/em_netlink.h
>> @@ -0,0 +1,16 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + *
>> + * Generic netlink for energy model.
>> + *
>> + * Copyright (c) 2025 Valve Corporation.
>> + * Author: Changwoo Min <changwoo@igalia.com>
>> + */
>> +#ifndef _EM_NETLINK_H
>> +#define _EM_NETLINK_H
>> +
>> +#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
>> +#else
>> +#endif
>> +
>> +#endif /* _EM_NETLINK_H */
> 
> Actually, those declarations of functions from patch 3/10 can
> live in this header. We would avoid creating more local headers
> in such case.

  That makes sense to me.

> 
> Then the patch 3/10 would have to go after this patch when
> this header is introduced.
> 
> Please ignore the comment in the patch 3/10 and try to
> use this header. It is also logically linked to the
> notifications, so belongs to such header IMHO.
>

Sure, after moving 3/10 after this, I will move the changes made in 3/10
to em_netlink.h. I will keep the implementation of
for_each_em_perf_domain() and em_perf_domain_get_by_id() in
energy_model.c since it it not ideal to expose em_pd_list_mutex, etc.
outside of energy_mode.c. And, this requires to include "em_netlink.h"
from energy_model.c.

Regards,
Changwoo Min

Re: [PATCH RESEND v4 07/10] PM: EM: Implement em_nl_get_pd_table_doit()

[Changwoo Min]

On 10/10/25 19:01, Lukasz Luba wrote:
> 
> 
> On 9/21/25 04:19, Changwoo Min wrote:
>> When a userspace requests EM_CMD_GET_PD_TABLE with an ID of a 
>> performancei
> 
> s/performancei/performance/

Will fix it. Thanks!

[...]

>>   int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info 
>> *info)
>>   {
>> -    return -EOPNOTSUPP;
>> +    struct sk_buff *msg;
>> +    struct em_perf_domain *pd;
>> +    void *hdr;
>> +    int cmd = info->genlhdr->cmd;
>> +    int msg_sz, ret = -EMSGSIZE;
> 
> Please put them in the reverse christmas tree order.
> 

Sure, will fix it as suggested in the next version.

Regards,
Changwoo Min

> 
> After fixing that cosmetic thing you can add:
> 
> Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
> 

Re: [PATCH RESEND v4 05/10] PM: EM: Add a skeleton code for netlink notification

[Lukasz Luba]

On 10/13/25 14:46, Changwoo Min wrote:
> 
> 
> On 10/7/25 00:44, Lukasz Luba wrote:
>>
>>
>> On 9/21/25 04:19, Changwoo Min wrote:
>>> Add a boilerplate code for netlink notification to register the new
>>> protocol family. Also, initialize and register the netlink during 
>>> booting.
>>> The initialization is called at the postcore level, which is late enough
>>> after the generic netlink is initialized.
>>>
>>> Finally, update MAINTAINERS to include new files.
>>>
>>> Signed-off-by: Changwoo Min <changwoo@igalia.com>
>>> ---
>>>   MAINTAINERS               |  2 +-
>>>   kernel/power/Makefile     |  5 ++++-
>>>   kernel/power/em_netlink.c | 35 +++++++++++++++++++++++++++++++++++
>>>   kernel/power/em_netlink.h | 16 ++++++++++++++++
>>>   4 files changed, 56 insertions(+), 2 deletions(-)
>>>   create mode 100644 kernel/power/em_netlink.c
>>>   create mode 100644 kernel/power/em_netlink.h
>>>
>>> diff --git a/MAINTAINERS b/MAINTAINERS
>>> index 0992029d271d..ba528836eac1 100644
>>> --- a/MAINTAINERS
>>> +++ b/MAINTAINERS
>>> @@ -9034,7 +9034,7 @@ F:    include/linux/energy_model.h
>>>   F:    Documentation/power/energy-model.rst
>>>   F:    Documentation/netlink/specs/em.yaml
>>>   F:    include/uapi/linux/energy_model.h
>>> -F:    kernel/power/em_netlink_autogen.*
>>> +F:    kernel/power/em_netlink*.*
>>>   EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
>>>   M:    Laurentiu Tudor <laurentiu.tudor@nxp.com>
>>> diff --git a/kernel/power/Makefile b/kernel/power/Makefile
>>> index 874ad834dc8d..284a760aade7 100644
>>> --- a/kernel/power/Makefile
>>> +++ b/kernel/power/Makefile
>>> @@ -21,4 +21,7 @@ obj-$(CONFIG_PM_WAKELOCKS)    += wakelock.o
>>>   obj-$(CONFIG_MAGIC_SYSRQ)    += poweroff.o
>>> -obj-$(CONFIG_ENERGY_MODEL)    += energy_model.o
>>> +obj-$(CONFIG_ENERGY_MODEL)    += em.o
>>> +em-y                := energy_model.o
>>> +em-$(CONFIG_NET)        += em_netlink_autogen.o em_netlink.o
>>> +
>>> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
>>> new file mode 100644
>>> index 000000000000..f3fbfeff29a4
>>> --- /dev/null
>>> +++ b/kernel/power/em_netlink.c
>>> @@ -0,0 +1,35 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +/*
>>> + *
>>> + * Generic netlink for energy model.
>>> + *
>>> + * Copyright (c) 2025 Valve Corporation.
>>> + * Author: Changwoo Min <changwoo@igalia.com>
>>> + */
>>> +
>>> +#define pr_fmt(fmt) "energy_model: " fmt
>>> +
>>> +#include <linux/energy_model.h>
>>> +#include <net/sock.h>
>>> +#include <net/genetlink.h>
>>> +#include <uapi/linux/energy_model.h>
>>> +
>>> +#include "em_netlink.h"
>>> +#include "em_netlink_autogen.h"
>>> +
>>> +int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
>>> +{
>>> +    return -EOPNOTSUPP;
>>> +}
>>> +
>>> +int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info 
>>> *info)
>>> +{
>>> +    return -EOPNOTSUPP;
>>> +}
>>> +
>>> +static int __init em_netlink_init(void)
>>> +{
>>> +    return genl_register_family(&em_nl_family);
>>> +}
>>> +postcore_initcall(em_netlink_init);
>>> +
>>> diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
>>> new file mode 100644
>>> index 000000000000..acd186c92d6b
>>> --- /dev/null
>>> +++ b/kernel/power/em_netlink.h
>>> @@ -0,0 +1,16 @@
>>> +/* SPDX-License-Identifier: GPL-2.0 */
>>> +/*
>>> + *
>>> + * Generic netlink for energy model.
>>> + *
>>> + * Copyright (c) 2025 Valve Corporation.
>>> + * Author: Changwoo Min <changwoo@igalia.com>
>>> + */
>>> +#ifndef _EM_NETLINK_H
>>> +#define _EM_NETLINK_H
>>> +
>>> +#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
>>> +#else
>>> +#endif
>>> +
>>> +#endif /* _EM_NETLINK_H */
>>
>> Actually, those declarations of functions from patch 3/10 can
>> live in this header. We would avoid creating more local headers
>> in such case.
> 
>   That makes sense to me.
> 
>>
>> Then the patch 3/10 would have to go after this patch when
>> this header is introduced.
>>
>> Please ignore the comment in the patch 3/10 and try to
>> use this header. It is also logically linked to the
>> notifications, so belongs to such header IMHO.
>>
> 
> Sure, after moving 3/10 after this, I will move the changes made in 3/10
> to em_netlink.h. I will keep the implementation of
> for_each_em_perf_domain() and em_perf_domain_get_by_id() in
> energy_model.c since it it not ideal to expose em_pd_list_mutex, etc.
> outside of energy_mode.c. And, this requires to include "em_netlink.h"
> from energy_model.c.
> 

Sounds good. These are only minor changes because it has nice shape
already.

I was able to test it last weekend and it works on my setup now.
I can see the messages coming in the user-space, so it should
work as designed.

Thank you Changwoo! Looking forward for your v5.

Regards,
Lukasz

Re: [PATCH RESEND v4 05/10] PM: EM: Add a skeleton code for netlink notification

[Changwoo Min]

On 10/13/25 22:53, Lukasz Luba wrote:
> 
> 
> On 10/13/25 14:46, Changwoo Min wrote:
>>
>>
>> On 10/7/25 00:44, Lukasz Luba wrote:
>>>
>>>
>>> On 9/21/25 04:19, Changwoo Min wrote:
>>>> Add a boilerplate code for netlink notification to register the new
>>>> protocol family. Also, initialize and register the netlink during 
>>>> booting.
>>>> The initialization is called at the postcore level, which is late 
>>>> enough
>>>> after the generic netlink is initialized.
>>>>
>>>> Finally, update MAINTAINERS to include new files.
>>>>
>>>> Signed-off-by: Changwoo Min <changwoo@igalia.com>
>>>> ---
>>>>   MAINTAINERS               |  2 +-
>>>>   kernel/power/Makefile     |  5 ++++-
>>>>   kernel/power/em_netlink.c | 35 +++++++++++++++++++++++++++++++++++
>>>>   kernel/power/em_netlink.h | 16 ++++++++++++++++
>>>>   4 files changed, 56 insertions(+), 2 deletions(-)
>>>>   create mode 100644 kernel/power/em_netlink.c
>>>>   create mode 100644 kernel/power/em_netlink.h
>>>>
>>>> diff --git a/MAINTAINERS b/MAINTAINERS
>>>> index 0992029d271d..ba528836eac1 100644
>>>> --- a/MAINTAINERS
>>>> +++ b/MAINTAINERS
>>>> @@ -9034,7 +9034,7 @@ F:    include/linux/energy_model.h
>>>>   F:    Documentation/power/energy-model.rst
>>>>   F:    Documentation/netlink/specs/em.yaml
>>>>   F:    include/uapi/linux/energy_model.h
>>>> -F:    kernel/power/em_netlink_autogen.*
>>>> +F:    kernel/power/em_netlink*.*
>>>>   EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
>>>>   M:    Laurentiu Tudor <laurentiu.tudor@nxp.com>
>>>> diff --git a/kernel/power/Makefile b/kernel/power/Makefile
>>>> index 874ad834dc8d..284a760aade7 100644
>>>> --- a/kernel/power/Makefile
>>>> +++ b/kernel/power/Makefile
>>>> @@ -21,4 +21,7 @@ obj-$(CONFIG_PM_WAKELOCKS)    += wakelock.o
>>>>   obj-$(CONFIG_MAGIC_SYSRQ)    += poweroff.o
>>>> -obj-$(CONFIG_ENERGY_MODEL)    += energy_model.o
>>>> +obj-$(CONFIG_ENERGY_MODEL)    += em.o
>>>> +em-y                := energy_model.o
>>>> +em-$(CONFIG_NET)        += em_netlink_autogen.o em_netlink.o
>>>> +
>>>> diff --git a/kernel/power/em_netlink.c b/kernel/power/em_netlink.c
>>>> new file mode 100644
>>>> index 000000000000..f3fbfeff29a4
>>>> --- /dev/null
>>>> +++ b/kernel/power/em_netlink.c
>>>> @@ -0,0 +1,35 @@
>>>> +// SPDX-License-Identifier: GPL-2.0
>>>> +/*
>>>> + *
>>>> + * Generic netlink for energy model.
>>>> + *
>>>> + * Copyright (c) 2025 Valve Corporation.
>>>> + * Author: Changwoo Min <changwoo@igalia.com>
>>>> + */
>>>> +
>>>> +#define pr_fmt(fmt) "energy_model: " fmt
>>>> +
>>>> +#include <linux/energy_model.h>
>>>> +#include <net/sock.h>
>>>> +#include <net/genetlink.h>
>>>> +#include <uapi/linux/energy_model.h>
>>>> +
>>>> +#include "em_netlink.h"
>>>> +#include "em_netlink_autogen.h"
>>>> +
>>>> +int em_nl_get_pds_doit(struct sk_buff *skb, struct genl_info *info)
>>>> +{
>>>> +    return -EOPNOTSUPP;
>>>> +}
>>>> +
>>>> +int em_nl_get_pd_table_doit(struct sk_buff *skb, struct genl_info 
>>>> *info)
>>>> +{
>>>> +    return -EOPNOTSUPP;
>>>> +}
>>>> +
>>>> +static int __init em_netlink_init(void)
>>>> +{
>>>> +    return genl_register_family(&em_nl_family);
>>>> +}
>>>> +postcore_initcall(em_netlink_init);
>>>> +
>>>> diff --git a/kernel/power/em_netlink.h b/kernel/power/em_netlink.h
>>>> new file mode 100644
>>>> index 000000000000..acd186c92d6b
>>>> --- /dev/null
>>>> +++ b/kernel/power/em_netlink.h
>>>> @@ -0,0 +1,16 @@
>>>> +/* SPDX-License-Identifier: GPL-2.0 */
>>>> +/*
>>>> + *
>>>> + * Generic netlink for energy model.
>>>> + *
>>>> + * Copyright (c) 2025 Valve Corporation.
>>>> + * Author: Changwoo Min <changwoo@igalia.com>
>>>> + */
>>>> +#ifndef _EM_NETLINK_H
>>>> +#define _EM_NETLINK_H
>>>> +
>>>> +#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_NET)
>>>> +#else
>>>> +#endif
>>>> +
>>>> +#endif /* _EM_NETLINK_H */
>>>
>>> Actually, those declarations of functions from patch 3/10 can
>>> live in this header. We would avoid creating more local headers
>>> in such case.
>>
>>   That makes sense to me.
>>
>>>
>>> Then the patch 3/10 would have to go after this patch when
>>> this header is introduced.
>>>
>>> Please ignore the comment in the patch 3/10 and try to
>>> use this header. It is also logically linked to the
>>> notifications, so belongs to such header IMHO.
>>>
>>
>> Sure, after moving 3/10 after this, I will move the changes made in 3/10
>> to em_netlink.h. I will keep the implementation of
>> for_each_em_perf_domain() and em_perf_domain_get_by_id() in
>> energy_model.c since it it not ideal to expose em_pd_list_mutex, etc.
>> outside of energy_mode.c. And, this requires to include "em_netlink.h"
>> from energy_model.c.
>>
> 
> Sounds good. These are only minor changes because it has nice shape
> already.
> 
> I was able to test it last weekend and it works on my setup now.
> I can see the messages coming in the user-space, so it should
> work as designed.
> 
> Thank you Changwoo! Looking forward for your v5.

Thank you Lukasz for testing. I will send v5 soon after final testing.

Regards,
Changwoo Min