[PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Hi!

These are the power sequencing patches sent separately after some
improvements suggested by Bjorn Helgaas. I intend to pick them up into a
new branch and maintain the subsystem from now on. I then plan to
provide an immutable tag to the Bluetooth and PCI subsystems so that the
rest of the C changes can be applied. This new branch will then be
directly sent to Linus Torvalds for the next merge window.

Changelog:

Since v8:
- split the pwrseq patches out into their own series
- rename incref/decref functions to get/put for consistency
- fix typos
- make it very explicit in docs that arrays of targets and units must be
  NULL-terminated
- Link to v8: https://lore.kernel.org/r/20240528-pwrseq-v8-0-d354d52b763c@linaro.org

Since v7:
- added DTS changes for sm8650-hdk
- added circular dependency detection for pwrseq units
- fixed a KASAN reported use-after-free error in remove path
- improve Kconfig descriptions
- fix typos in bindings and Kconfig
- fixed issues reported by smatch
- fix the unbind path in PCI pwrctl
- lots of minor improvements to the pwrseq core

Since v6:
- kernel doc fixes
- drop myself from the DT bindings maintainers list for ath12k
- wait until the PCI bridge device is fully added before creating the
  PCI pwrctl platform devices for its sub-nodes, otherwise we may see
  sysfs and procfs attribute failures (due to duplication, we're
  basically trying to probe the same device twice at the same time)
- I kept the regulators for QCA6390's ath11k as required as they only
  apply to this specific Qualcomm package

Since v5:
- unify the approach to modelling the WCN WLAN/BT chips by always exposing
  the PMU node on the device tree and making the WLAN and BT nodes become
  consumers of its power outputs; this includes a major rework of the DT
  sources, bindings and driver code; there's no more a separate PCI
  pwrctl driver for WCN7850, instead its power-up sequence was moved
  into the pwrseq driver common for all WCN chips
- don't set load_uA from new regulator consumers
- fix reported kerneldoc issues
- drop voltage ranges for PMU outputs from DT
- many minor tweaks and reworks

v1: Original RFC:

https://lore.kernel.org/lkml/20240104130123.37115-1-brgl@bgdev.pl/T/

v2: First real patch series (should have been PATCH v2) adding what I
    referred to back then as PCI power sequencing:

https://lore.kernel.org/linux-arm-kernel/2024021413-grumbling-unlivable-c145@gregkh/T/

v3: RFC for the DT representation of the PMU supplying the WLAN and BT
    modules inside the QCA6391 package (was largely separate from the
    series but probably should have been called PATCH or RFC v3):

https://lore.kernel.org/all/CAMRc=Mc+GNoi57eTQg71DXkQKjdaoAmCpB=h2ndEpGnmdhVV-Q@mail.gmail.com/T/

v4: Second attempt at the full series with changed scope (introduction of
    the pwrseq subsystem, should have been RFC v4)

https://lore.kernel.org/lkml/20240201155532.49707-1-brgl@bgdev.pl/T/

v5: Two different ways of handling QCA6390 and WCN7850:

https://lore.kernel.org/lkml/20240216203215.40870-1-brgl@bgdev.pl/

Bartosz Golaszewski (2):
  power: sequencing: implement the pwrseq core
  power: pwrseq: add a driver for the PMU module on the QCom WCN
    chipsets

 MAINTAINERS                                |    8 +
 drivers/power/Kconfig                      |    1 +
 drivers/power/Makefile                     |    1 +
 drivers/power/sequencing/Kconfig           |   29 +
 drivers/power/sequencing/Makefile          |    6 +
 drivers/power/sequencing/core.c            | 1105 ++++++++++++++++++++
 drivers/power/sequencing/pwrseq-qcom-wcn.c |  336 ++++++
 include/linux/pwrseq/consumer.h            |   56 +
 include/linux/pwrseq/provider.h            |   75 ++
 9 files changed, 1617 insertions(+)
 create mode 100644 drivers/power/sequencing/Kconfig
 create mode 100644 drivers/power/sequencing/Makefile
 create mode 100644 drivers/power/sequencing/core.c
 create mode 100644 drivers/power/sequencing/pwrseq-qcom-wcn.c
 create mode 100644 include/linux/pwrseq/consumer.h
 create mode 100644 include/linux/pwrseq/provider.h

-- 
2.40.1

[PATCH v9 1/2] power: sequencing: implement the pwrseq core

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Implement the power sequencing subsystem allowing devices to share
complex powering-up and down procedures. It's split into the consumer
and provider parts but does not implement any new DT bindings so that
the actual power sequencing is never revealed in the DT representation.

Tested-by: Amit Pundir <amit.pundir@linaro.org>
Tested-by: Neil Armstrong <neil.armstrong@linaro.org> # on SM8550-QRD, SM8650-QRD & SM8650-HDK
Tested-by: Caleb Connolly <caleb.connolly@linaro.org> # OnePlus 8T
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 MAINTAINERS                       |    8 +
 drivers/power/Kconfig             |    1 +
 drivers/power/Makefile            |    1 +
 drivers/power/sequencing/Kconfig  |   12 +
 drivers/power/sequencing/Makefile |    4 +
 drivers/power/sequencing/core.c   | 1105 +++++++++++++++++++++++++++++
 include/linux/pwrseq/consumer.h   |   56 ++
 include/linux/pwrseq/provider.h   |   75 ++
 8 files changed, 1262 insertions(+)
 create mode 100644 drivers/power/sequencing/Kconfig
 create mode 100644 drivers/power/sequencing/Makefile
 create mode 100644 drivers/power/sequencing/core.c
 create mode 100644 include/linux/pwrseq/consumer.h
 create mode 100644 include/linux/pwrseq/provider.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 242fc612fbc5..79d118779121 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -17931,6 +17931,14 @@ F:	include/linux/pm_*
 F:	include/linux/powercap.h
 F:	kernel/configs/nopm.config
 
+POWER SEQUENCING
+M:	Bartosz Golaszewski <brgl@bgdev.pl>
+L:	linux-pm@vger.kernel.org
+S:	Maintained
+T:	git git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux.git
+F:	drivers/power/sequencing/
+F:	include/linux/pwrseq/
+
 POWER STATE COORDINATION INTERFACE (PSCI)
 M:	Mark Rutland <mark.rutland@arm.com>
 M:	Lorenzo Pieralisi <lpieralisi@kernel.org>
diff --git a/drivers/power/Kconfig b/drivers/power/Kconfig
index 696bf77a7042..9a8e44ca9ae4 100644
--- a/drivers/power/Kconfig
+++ b/drivers/power/Kconfig
@@ -1,3 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0-only
 source "drivers/power/reset/Kconfig"
+source "drivers/power/sequencing/Kconfig"
 source "drivers/power/supply/Kconfig"
diff --git a/drivers/power/Makefile b/drivers/power/Makefile
index effbf0377f32..962a2cd30a51 100644
--- a/drivers/power/Makefile
+++ b/drivers/power/Makefile
@@ -1,3 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0-only
 obj-$(CONFIG_POWER_RESET)	+= reset/
+obj-$(CONFIG_POWER_SEQUENCING)	+= sequencing/
 obj-$(CONFIG_POWER_SUPPLY)	+= supply/
diff --git a/drivers/power/sequencing/Kconfig b/drivers/power/sequencing/Kconfig
new file mode 100644
index 000000000000..ba5732b1dbf8
--- /dev/null
+++ b/drivers/power/sequencing/Kconfig
@@ -0,0 +1,12 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+menuconfig POWER_SEQUENCING
+	tristate "Power Sequencing support"
+	help
+	  Say Y here to enable the Power Sequencing subsystem.
+
+	  This subsystem is designed to control power to devices that share
+	  complex resources and/or require specific power sequences to be run
+	  during power-up.
+
+	  If unsure, say no.
diff --git a/drivers/power/sequencing/Makefile b/drivers/power/sequencing/Makefile
new file mode 100644
index 000000000000..dcdf8c0c159e
--- /dev/null
+++ b/drivers/power/sequencing/Makefile
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0
+
+obj-$(CONFIG_POWER_SEQUENCING)		+= pwrseq-core.o
+pwrseq-core-y				:= core.o
diff --git a/drivers/power/sequencing/core.c b/drivers/power/sequencing/core.c
new file mode 100644
index 000000000000..09c0f27c8aa7
--- /dev/null
+++ b/drivers/power/sequencing/core.c
@@ -0,0 +1,1105 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2024 Linaro Ltd.
+ */
+
+#include <linux/bug.h>
+#include <linux/cleanup.h>
+#include <linux/debugfs.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/export.h>
+#include <linux/idr.h>
+#include <linux/kernel.h>
+#include <linux/kref.h>
+#include <linux/list.h>
+#include <linux/lockdep.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/property.h>
+#include <linux/pwrseq/consumer.h>
+#include <linux/pwrseq/provider.h>
+#include <linux/radix-tree.h>
+#include <linux/rwsem.h>
+#include <linux/slab.h>
+
+/*
+ * Power-sequencing framework for linux.
+ *
+ * This subsystem allows power sequence providers to register a set of targets
+ * that consumers may request and power-up/down.
+ *
+ * Glossary:
+ *
+ * Unit - a unit is a discreet chunk of a power sequence. For instance one unit
+ * may enable a set of regulators, another may enable a specific GPIO. Units
+ * can define dependencies in the form of other units that must be enabled
+ * before it itself can be.
+ *
+ * Target - a target is a set of units (composed of the "final" unit and its
+ * dependencies) that a consumer selects by its name when requesting a handle
+ * to the power sequencer. Via the dependency system, multiple targets may
+ * share the same parts of a power sequence but ignore parts that are
+ * irrelevant.
+ *
+ * Descriptor - a handle passed by the pwrseq core to every consumer that
+ * serves as the entry point to the provider layer. It ensures coherence
+ * between different users and keeps reference counting consistent.
+ *
+ * Each provider must define a .match() callback whose role is to determine
+ * whether a potential consumer is in fact associated with this sequencer.
+ * This allows creating abstraction layers on top of regular device-tree
+ * resources like regulators, clocks and other nodes connected to the consumer
+ * via phandle.
+ */
+
+static DEFINE_IDA(pwrseq_ida);
+
+/*
+ * Protects the device list on the pwrseq bus from concurrent modifications
+ * but allows simultaneous read-only access.
+ */
+static DECLARE_RWSEM(pwrseq_sem);
+
+/**
+ * struct pwrseq_unit - Private power-sequence unit data.
+ * @ref: Reference count for this object. When it goes to 0, the object is
+ *       destroyed.
+ * @name: Name of this target.
+ * @list: Link to siblings on the list of all units of a single sequencer.
+ * @deps: List of units on which this unit depends.
+ * @enable: Callback running the part of the power-on sequence provided by
+ *          this unit.
+ * @disable: Callback running the part of the power-off sequence provided
+ *           by this unit.
+ * @enable_count: Current number of users that enabled this unit. May be the
+ *                consumer of the power sequencer or other units that depend
+ *                on this one.
+ */
+struct pwrseq_unit {
+	struct kref ref;
+	const char *name;
+	struct list_head list;
+	struct list_head deps;
+	pwrseq_power_state_func enable;
+	pwrseq_power_state_func disable;
+	unsigned int enable_count;
+};
+
+static struct pwrseq_unit *pwrseq_unit_new(const struct pwrseq_unit_data *data)
+{
+	struct pwrseq_unit *unit;
+
+	unit = kzalloc(sizeof(*unit), GFP_KERNEL);
+	if (!unit)
+		return NULL;
+
+	unit->name = kstrdup_const(data->name, GFP_KERNEL);
+	if (!unit->name) {
+		kfree(unit);
+		return NULL;
+	}
+
+	kref_init(&unit->ref);
+	INIT_LIST_HEAD(&unit->deps);
+	unit->enable = data->enable;
+	unit->disable = data->disable;
+
+	return unit;
+}
+
+static struct pwrseq_unit *pwrseq_unit_get(struct pwrseq_unit *unit)
+{
+	kref_get(&unit->ref);
+
+	return unit;
+}
+
+static void pwrseq_unit_release(struct kref *ref);
+
+static void pwrseq_unit_put(struct pwrseq_unit *unit)
+{
+	kref_put(&unit->ref, pwrseq_unit_release);
+}
+
+/**
+ * struct pwrseq_unit_dep - Wrapper around a reference to the unit structure
+ *                          allowing to keep it on multiple dependency lists
+ *                          in different units.
+ * @list: Siblings on the list.
+ * @unit: Address of the referenced unit.
+ */
+struct pwrseq_unit_dep {
+	struct list_head list;
+	struct pwrseq_unit *unit;
+};
+
+static struct pwrseq_unit_dep *pwrseq_unit_dep_new(struct pwrseq_unit *unit)
+{
+	struct pwrseq_unit_dep *dep;
+
+	dep = kzalloc(sizeof(*dep), GFP_KERNEL);
+	if (!dep)
+		return NULL;
+
+	dep->unit = unit;
+
+	return dep;
+}
+
+static void pwrseq_unit_dep_free(struct pwrseq_unit_dep *ref)
+{
+	pwrseq_unit_put(ref->unit);
+	kfree(ref);
+}
+
+static void pwrseq_unit_free_deps(struct list_head *list)
+{
+	struct pwrseq_unit_dep *dep, *next;
+
+	list_for_each_entry_safe(dep, next, list, list) {
+		list_del(&dep->list);
+		pwrseq_unit_dep_free(dep);
+	}
+}
+
+static void pwrseq_unit_release(struct kref *ref)
+{
+	struct pwrseq_unit *unit = container_of(ref, struct pwrseq_unit, ref);
+
+	pwrseq_unit_free_deps(&unit->deps);
+	list_del(&unit->list);
+	kfree_const(unit->name);
+	kfree(unit);
+}
+
+/**
+ * struct pwrseq_target - Private power-sequence target data.
+ * @list: Siblings on the list of all targets exposed by a power sequencer.
+ * @name: Name of the target.
+ * @unit: Final unit for this target.
+ * @post_enable: Callback run after the target unit has been enabled, *after*
+ *               the state lock has been released. It's useful for implementing
+ *               boot-up delays without blocking other users from powering up
+ *               using the same power sequencer.
+ */
+struct pwrseq_target {
+	struct list_head list;
+	const char *name;
+	struct pwrseq_unit *unit;
+	pwrseq_power_state_func post_enable;
+};
+
+static struct pwrseq_target *
+pwrseq_target_new(const struct pwrseq_target_data *data)
+{
+	struct pwrseq_target *target;
+
+	target = kzalloc(sizeof(*target), GFP_KERNEL);
+	if (!target)
+		return NULL;
+
+	target->name = kstrdup_const(data->name, GFP_KERNEL);
+	if (!target->name) {
+		kfree(target);
+		return NULL;
+	}
+
+	target->post_enable = data->post_enable;
+
+	return target;
+}
+
+static void pwrseq_target_free(struct pwrseq_target *target)
+{
+	pwrseq_unit_put(target->unit);
+	kfree_const(target->name);
+	kfree(target);
+}
+
+/**
+ * struct pwrseq_device - Private power sequencing data.
+ * @dev: Device struct associated with this sequencer.
+ * @id: Device ID.
+ * @owner: Prevents removal of active power sequencing providers.
+ * @rw_lock: Protects the device from being unregistered while in use.
+ * @state_lock: Prevents multiple users running the power sequence at the same
+ *              time.
+ * @match: Power sequencer matching callback.
+ * @targets: List of targets exposed by this sequencer.
+ * @units: List of all units supported by this sequencer.
+ */
+struct pwrseq_device {
+	struct device dev;
+	int id;
+	struct module *owner;
+	struct rw_semaphore rw_lock;
+	struct mutex state_lock;
+	pwrseq_match_func match;
+	struct list_head targets;
+	struct list_head units;
+};
+
+static struct pwrseq_device *to_pwrseq_device(struct device *dev)
+{
+	return container_of(dev, struct pwrseq_device, dev);
+}
+
+static struct pwrseq_device *pwrseq_device_get(struct pwrseq_device *pwrseq)
+{
+	get_device(&pwrseq->dev);
+
+	return pwrseq;
+}
+
+static void pwrseq_device_put(struct pwrseq_device *pwrseq)
+{
+	put_device(&pwrseq->dev);
+}
+
+/**
+ * struct pwrseq_desc - Wraps access to the pwrseq_device and ensures that one
+ *                      user cannot break the reference counting for others.
+ * @pwrseq: Reference to the power sequencing device.
+ * @target: Reference to the target this descriptor allows to control.
+ * @powered_on: Power state set by the holder of the descriptor (not necessarily
+ * corresponding to the actual power state of the device).
+ */
+struct pwrseq_desc {
+	struct pwrseq_device *pwrseq;
+	struct pwrseq_target *target;
+	bool powered_on;
+};
+
+static const struct bus_type pwrseq_bus = {
+	.name = "pwrseq",
+};
+
+static void pwrseq_release(struct device *dev)
+{
+	struct pwrseq_device *pwrseq = to_pwrseq_device(dev);
+	struct pwrseq_target *target, *pos;
+
+	list_for_each_entry_safe(target, pos, &pwrseq->targets, list) {
+		list_del(&target->list);
+		pwrseq_target_free(target);
+	}
+
+	mutex_destroy(&pwrseq->state_lock);
+	ida_free(&pwrseq_ida, pwrseq->id);
+	kfree(pwrseq);
+}
+
+static const struct device_type pwrseq_device_type = {
+	.name = "power_sequencer",
+	.release = pwrseq_release,
+};
+
+static int pwrseq_check_unit_deps(const struct pwrseq_unit_data *data,
+				  struct radix_tree_root *visited_units)
+{
+	const struct pwrseq_unit_data *tmp, **cur;
+	int ret;
+
+	ret = radix_tree_insert(visited_units, (unsigned long)data,
+				(void *)data);
+	if (ret)
+		return ret;
+
+	for (cur = data->deps; cur && *cur; cur++) {
+		tmp = radix_tree_lookup(visited_units, (unsigned long)*cur);
+		if (tmp) {
+			WARN(1, "Circular dependency in power sequencing flow detected!\n");
+			return -EINVAL;
+		}
+
+		ret = pwrseq_check_unit_deps(*cur, visited_units);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int pwrseq_check_target_deps(const struct pwrseq_target_data *data)
+{
+	struct radix_tree_root visited_units;
+	struct radix_tree_iter iter;
+	void __rcu **slot;
+	int ret;
+
+	if (!data->unit)
+		return -EINVAL;
+
+	INIT_RADIX_TREE(&visited_units, GFP_KERNEL);
+	ret = pwrseq_check_unit_deps(data->unit, &visited_units);
+	radix_tree_for_each_slot(slot, &visited_units, &iter, 0)
+		radix_tree_delete(&visited_units, iter.index);
+
+	return ret;
+}
+
+static int pwrseq_unit_setup_deps(const struct pwrseq_unit_data **data,
+				  struct list_head *dep_list,
+				  struct list_head *unit_list,
+				  struct radix_tree_root *processed_units);
+
+static struct pwrseq_unit *
+pwrseq_unit_setup(const struct pwrseq_unit_data *data,
+		  struct list_head *unit_list,
+		  struct radix_tree_root *processed_units)
+{
+	struct pwrseq_unit *unit;
+	int ret;
+
+	unit = radix_tree_lookup(processed_units, (unsigned long)data);
+	if (unit)
+		return pwrseq_unit_get(unit);
+
+	unit = pwrseq_unit_new(data);
+	if (!unit)
+		return ERR_PTR(-ENOMEM);
+
+	if (data->deps) {
+		ret = pwrseq_unit_setup_deps(data->deps, &unit->deps,
+					     unit_list, processed_units);
+		if (ret) {
+			pwrseq_unit_put(unit);
+			return ERR_PTR(ret);
+		}
+	}
+
+	ret = radix_tree_insert(processed_units, (unsigned long)data, unit);
+	if (ret) {
+		pwrseq_unit_put(unit);
+		return ERR_PTR(ret);
+	}
+
+	list_add_tail(&unit->list, unit_list);
+
+	return unit;
+}
+
+static int pwrseq_unit_setup_deps(const struct pwrseq_unit_data **data,
+				  struct list_head *dep_list,
+				  struct list_head *unit_list,
+				  struct radix_tree_root *processed_units)
+{
+	const struct pwrseq_unit_data *pos;
+	struct pwrseq_unit_dep *dep;
+	struct pwrseq_unit *unit;
+	int i;
+
+	for (i = 0; data[i]; i++) {
+		pos = data[i];
+
+		unit = pwrseq_unit_setup(pos, unit_list, processed_units);
+		if (IS_ERR(unit))
+			return PTR_ERR(unit);
+
+		dep = pwrseq_unit_dep_new(unit);
+		if (!dep) {
+			pwrseq_unit_put(unit);
+			return -ENOMEM;
+		}
+
+		list_add_tail(&dep->list, dep_list);
+	}
+
+	return 0;
+}
+
+static int pwrseq_do_setup_targets(const struct pwrseq_target_data **data,
+				   struct pwrseq_device *pwrseq,
+				   struct radix_tree_root *processed_units)
+{
+	const struct pwrseq_target_data *pos;
+	struct pwrseq_target *target;
+	int ret, i;
+
+	for (i = 0; data[i]; i++) {
+		pos = data[i];
+
+		ret = pwrseq_check_target_deps(pos);
+		if (ret)
+			return ret;
+
+		target = pwrseq_target_new(pos);
+		if (!target)
+			return -ENOMEM;
+
+		target->unit = pwrseq_unit_setup(pos->unit, &pwrseq->units,
+						 processed_units);
+		if (IS_ERR(target->unit)) {
+			ret = PTR_ERR(target->unit);
+			pwrseq_target_free(target);
+			return ret;
+		}
+
+		list_add_tail(&target->list, &pwrseq->targets);
+	}
+
+	return 0;
+}
+
+static int pwrseq_setup_targets(const struct pwrseq_target_data **targets,
+				struct pwrseq_device *pwrseq)
+{
+	struct radix_tree_root processed_units;
+	struct radix_tree_iter iter;
+	void __rcu **slot;
+	int ret;
+
+	INIT_RADIX_TREE(&processed_units, GFP_KERNEL);
+	ret = pwrseq_do_setup_targets(targets, pwrseq, &processed_units);
+	radix_tree_for_each_slot(slot, &processed_units, &iter, 0)
+		radix_tree_delete(&processed_units, iter.index);
+
+	return ret;
+}
+
+/**
+ * pwrseq_device_register() - Register a new power sequencer.
+ * @config: Configuration of the new power sequencing device.
+ *
+ * The config structure is only used during the call and can be freed after
+ * the function returns. The config structure *must* have the parent device
+ * as well as the match() callback and at least one target set.
+ *
+ * Returns:
+ * Returns the address of the new pwrseq device or ERR_PTR() on failure.
+ */
+struct pwrseq_device *
+pwrseq_device_register(const struct pwrseq_config *config)
+{
+	struct pwrseq_device *pwrseq;
+	int ret, id;
+
+	if (!config->parent || !config->match || !config->targets ||
+	    !config->targets[0])
+		return ERR_PTR(-EINVAL);
+
+	pwrseq = kzalloc(sizeof(*pwrseq), GFP_KERNEL);
+	if (!pwrseq)
+		return ERR_PTR(-ENOMEM);
+
+	pwrseq->dev.type = &pwrseq_device_type;
+	pwrseq->dev.bus = &pwrseq_bus;
+	pwrseq->dev.parent = config->parent;
+	device_set_node(&pwrseq->dev, dev_fwnode(config->parent));
+	dev_set_drvdata(&pwrseq->dev, config->drvdata);
+
+	id = ida_alloc(&pwrseq_ida, GFP_KERNEL);
+	if (id < 0) {
+		kfree(pwrseq);
+		return ERR_PTR(id);
+	}
+
+	pwrseq->id = id;
+
+	/*
+	 * From this point onwards the device's release() callback is
+	 * responsible for freeing resources.
+	 */
+	device_initialize(&pwrseq->dev);
+
+	ret = dev_set_name(&pwrseq->dev, "pwrseq.%d", pwrseq->id);
+	if (ret)
+		goto err_put_pwrseq;
+
+	pwrseq->owner = config->owner ?: THIS_MODULE;
+	pwrseq->match = config->match;
+
+	init_rwsem(&pwrseq->rw_lock);
+	mutex_init(&pwrseq->state_lock);
+	INIT_LIST_HEAD(&pwrseq->targets);
+	INIT_LIST_HEAD(&pwrseq->units);
+
+	ret = pwrseq_setup_targets(config->targets, pwrseq);
+	if (ret)
+		goto err_put_pwrseq;
+
+	scoped_guard(rwsem_write, &pwrseq_sem) {
+		ret = device_add(&pwrseq->dev);
+		if (ret)
+			goto err_put_pwrseq;
+	}
+
+	return pwrseq;
+
+err_put_pwrseq:
+	pwrseq_device_put(pwrseq);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL_GPL(pwrseq_device_register);
+
+/**
+ * pwrseq_device_unregister() - Unregister the power sequencer.
+ * @pwrseq: Power sequencer to unregister.
+ */
+void pwrseq_device_unregister(struct pwrseq_device *pwrseq)
+{
+	struct device *dev = &pwrseq->dev;
+	struct pwrseq_target *target;
+
+	scoped_guard(mutex, &pwrseq->state_lock) {
+		guard(rwsem_write)(&pwrseq->rw_lock);
+
+		list_for_each_entry(target, &pwrseq->targets, list)
+			WARN_ONCE(target->unit->enable_count,
+				  "REMOVING POWER SEQUENCER WITH ACTIVE USERS\n");
+
+		guard(rwsem_write)(&pwrseq_sem);
+
+		device_del(dev);
+	}
+
+	pwrseq_device_put(pwrseq);
+}
+EXPORT_SYMBOL_GPL(pwrseq_device_unregister);
+
+static void devm_pwrseq_device_unregister(void *data)
+{
+	struct pwrseq_device *pwrseq = data;
+
+	pwrseq_device_unregister(pwrseq);
+}
+
+/**
+ * devm_pwrseq_device_register() - Managed variant of pwrseq_device_register().
+ * @dev: Managing device.
+ * @config: Configuration of the new power sequencing device.
+ *
+ * Returns:
+ * Returns the address of the new pwrseq device or ERR_PTR() on failure.
+ */
+struct pwrseq_device *
+devm_pwrseq_device_register(struct device *dev,
+			    const struct pwrseq_config *config)
+{
+	struct pwrseq_device *pwrseq;
+	int ret;
+
+	pwrseq = pwrseq_device_register(config);
+	if (IS_ERR(pwrseq))
+		return pwrseq;
+
+	ret = devm_add_action_or_reset(dev, devm_pwrseq_device_unregister,
+				       pwrseq);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return pwrseq;
+}
+EXPORT_SYMBOL_GPL(devm_pwrseq_device_register);
+
+/**
+ * pwrseq_device_get_drvdata() - Get the driver private data associated with
+ *                               this sequencer.
+ * @pwrseq: Power sequencer object.
+ *
+ * Returns:
+ * Address of the private driver data.
+ */
+void *pwrseq_device_get_drvdata(struct pwrseq_device *pwrseq)
+{
+	return dev_get_drvdata(&pwrseq->dev);
+}
+EXPORT_SYMBOL_GPL(pwrseq_device_get_drvdata);
+
+struct pwrseq_match_data {
+	struct pwrseq_desc *desc;
+	struct device *dev;
+	const char *target;
+};
+
+static int pwrseq_match_device(struct device *pwrseq_dev, void *data)
+{
+	struct pwrseq_device *pwrseq = to_pwrseq_device(pwrseq_dev);
+	struct pwrseq_match_data *match_data = data;
+	struct pwrseq_target *target;
+	int ret;
+
+	lockdep_assert_held_read(&pwrseq_sem);
+
+	guard(rwsem_read)(&pwrseq->rw_lock);
+	if (!device_is_registered(&pwrseq->dev))
+		return 0;
+
+	ret = pwrseq->match(pwrseq, match_data->dev);
+	if (ret <= 0)
+		return ret;
+
+	/* We got the matching device, let's find the right target. */
+	list_for_each_entry(target, &pwrseq->targets, list) {
+		if (strcmp(target->name, match_data->target))
+			continue;
+
+		match_data->desc->target = target;
+	}
+
+	/*
+	 * This device does not have this target. No point in deferring as it
+	 * will not get a new target dynamically later.
+	 */
+	if (!match_data->desc->target)
+		return -ENOENT;
+
+	if (!try_module_get(pwrseq->owner))
+		return -EPROBE_DEFER;
+
+	match_data->desc->pwrseq = pwrseq_device_get(pwrseq);
+
+	return 1;
+}
+
+/**
+ * pwrseq_get() - Get the power sequencer associated with this device.
+ * @dev: Device for which to get the sequencer.
+ * @target: Name of the target exposed by the sequencer this device wants to
+ *          reach.
+ *
+ * Returns:
+ * New power sequencer descriptor for use by the consumer driver or ERR_PTR()
+ * on failure.
+ */
+struct pwrseq_desc *pwrseq_get(struct device *dev, const char *target)
+{
+	struct pwrseq_match_data match_data;
+	int ret;
+
+	struct pwrseq_desc *desc __free(kfree) = kzalloc(sizeof(*desc),
+							 GFP_KERNEL);
+	if (!desc)
+		return ERR_PTR(-ENOMEM);
+
+	match_data.desc = desc;
+	match_data.dev = dev;
+	match_data.target = target;
+
+	guard(rwsem_read)(&pwrseq_sem);
+
+	ret = bus_for_each_dev(&pwrseq_bus, NULL, &match_data,
+			       pwrseq_match_device);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	if (ret == 0)
+		/* No device matched. */
+		return ERR_PTR(-EPROBE_DEFER);
+
+	return no_free_ptr(desc);
+}
+EXPORT_SYMBOL_GPL(pwrseq_get);
+
+/**
+ * pwrseq_put() - Release the power sequencer descriptor.
+ * @desc: Descriptor to release.
+ */
+void pwrseq_put(struct pwrseq_desc *desc)
+{
+	struct pwrseq_device *pwrseq;
+
+	if (!desc)
+		return;
+
+	pwrseq = desc->pwrseq;
+
+	if (desc->powered_on)
+		pwrseq_power_off(desc);
+
+	kfree(desc);
+	module_put(pwrseq->owner);
+	pwrseq_device_put(pwrseq);
+}
+EXPORT_SYMBOL_GPL(pwrseq_put);
+
+static void devm_pwrseq_put(void *data)
+{
+	struct pwrseq_desc *desc = data;
+
+	pwrseq_put(desc);
+}
+
+/**
+ * devm_pwrseq_get() - Managed variant of pwrseq_get().
+ * @dev: Device for which to get the sequencer and which also manages its
+ *       lifetime.
+ * @target: Name of the target exposed by the sequencer this device wants to
+ *          reach.
+ *
+ * Returns:
+ * New power sequencer descriptor for use by the consumer driver or ERR_PTR()
+ * on failure.
+ */
+struct pwrseq_desc *devm_pwrseq_get(struct device *dev, const char *target)
+{
+	struct pwrseq_desc *desc;
+	int ret;
+
+	desc = pwrseq_get(dev, target);
+	if (IS_ERR(desc))
+		return desc;
+
+	ret = devm_add_action_or_reset(dev, devm_pwrseq_put, desc);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return desc;
+}
+EXPORT_SYMBOL_GPL(devm_pwrseq_get);
+
+static int pwrseq_unit_enable(struct pwrseq_device *pwrseq,
+			      struct pwrseq_unit *target);
+static int pwrseq_unit_disable(struct pwrseq_device *pwrseq,
+			       struct pwrseq_unit *target);
+
+static int pwrseq_unit_enable_deps(struct pwrseq_device *pwrseq,
+				   struct list_head *list)
+{
+	struct pwrseq_unit_dep *pos;
+	int ret = 0;
+
+	list_for_each_entry(pos, list, list) {
+		ret = pwrseq_unit_enable(pwrseq, pos->unit);
+		if (ret) {
+			list_for_each_entry_continue_reverse(pos, list, list)
+				pwrseq_unit_disable(pwrseq, pos->unit);
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static int pwrseq_unit_disable_deps(struct pwrseq_device *pwrseq,
+				    struct list_head *list)
+{
+	struct pwrseq_unit_dep *pos;
+	int ret = 0;
+
+	list_for_each_entry_reverse(pos, list, list) {
+		ret = pwrseq_unit_disable(pwrseq, pos->unit);
+		if (ret) {
+			list_for_each_entry_continue(pos, list, list)
+				pwrseq_unit_enable(pwrseq, pos->unit);
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static int pwrseq_unit_enable(struct pwrseq_device *pwrseq,
+			      struct pwrseq_unit *unit)
+{
+	int ret;
+
+	lockdep_assert_held_read(&pwrseq->rw_lock);
+	lockdep_assert_held(&pwrseq->state_lock);
+
+	if (unit->enable_count != 0) {
+		unit->enable_count++;
+		return 0;
+	}
+
+	ret = pwrseq_unit_enable_deps(pwrseq, &unit->deps);
+	if (ret) {
+		dev_err(&pwrseq->dev,
+			"Failed to enable dependencies before power-on for target '%s': %d\n",
+			unit->name, ret);
+		return ret;
+	}
+
+	if (unit->enable) {
+		ret = unit->enable(pwrseq);
+		if (ret) {
+			dev_err(&pwrseq->dev,
+				"Failed to enable target '%s': %d\n",
+				unit->name, ret);
+			pwrseq_unit_disable_deps(pwrseq, &unit->deps);
+			return ret;
+		}
+	}
+
+	unit->enable_count++;
+
+	return 0;
+}
+
+static int pwrseq_unit_disable(struct pwrseq_device *pwrseq,
+			       struct pwrseq_unit *unit)
+{
+	int ret;
+
+	lockdep_assert_held_read(&pwrseq->rw_lock);
+	lockdep_assert_held(&pwrseq->state_lock);
+
+	if (unit->enable_count == 0) {
+		WARN_ONCE(1, "Unmatched power-off for target '%s'\n",
+			  unit->name);
+		return -EBUSY;
+	}
+
+	if (unit->enable_count != 1) {
+		unit->enable_count--;
+		return 0;
+	}
+
+	if (unit->disable) {
+		ret = unit->disable(pwrseq);
+		if (ret) {
+			dev_err(&pwrseq->dev,
+				"Failed to disable target '%s': %d\n",
+				unit->name, ret);
+			return ret;
+		}
+	}
+
+	ret = pwrseq_unit_disable_deps(pwrseq, &unit->deps);
+	if (ret) {
+		dev_err(&pwrseq->dev,
+			"Failed to disable dependencies after power-off for target '%s': %d\n",
+			unit->name, ret);
+		if (unit->enable)
+			unit->enable(pwrseq);
+		return ret;
+	}
+
+	unit->enable_count--;
+
+	return 0;
+}
+
+/**
+ * pwrseq_power_on() - Issue a power-on request on behalf of the consumer
+ *                     device.
+ * @desc: Descriptor referencing the power sequencer.
+ *
+ * This function tells the power sequencer that the consumer wants to be
+ * powered-up. The sequencer may already have powered-up the device in which
+ * case the function returns 0. If the power-up sequence is already in
+ * progress, the function will block until it's done and return 0. If this is
+ * the first request, the device will be powered up.
+ *
+ * Returns:
+ * 0 on success, negative error number on failure.
+ */
+int pwrseq_power_on(struct pwrseq_desc *desc)
+{
+	struct pwrseq_device *pwrseq;
+	struct pwrseq_target *target;
+	struct pwrseq_unit *unit;
+	int ret;
+
+	might_sleep();
+
+	if (!desc || desc->powered_on)
+		return 0;
+
+	pwrseq = desc->pwrseq;
+	target = desc->target;
+	unit = target->unit;
+
+	guard(rwsem_read)(&pwrseq->rw_lock);
+	if (!device_is_registered(&pwrseq->dev))
+		return -ENODEV;
+
+	scoped_guard(mutex, &pwrseq->state_lock) {
+		ret = pwrseq_unit_enable(pwrseq, unit);
+		if (!ret)
+			desc->powered_on = true;
+	}
+
+	if (target->post_enable) {
+		ret = target->post_enable(pwrseq);
+		if (ret) {
+			pwrseq_unit_disable(pwrseq, unit);
+			desc->powered_on = false;
+		}
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(pwrseq_power_on);
+
+/**
+ * pwrseq_power_off() - Issue a power-off request on behalf of the consumer
+ *                      device.
+ * @desc: Descriptor referencing the power sequencer.
+ *
+ * This undoes the effects of pwrseq_power_on(). It issues a power-off request
+ * on behalf of the consumer and when the last remaining user does so, the
+ * power-down sequence will be started. If one is in progress, the function
+ * will block until it's complete and then return.
+ *
+ * Returns:
+ * 0 on success, negative error number on failure.
+ */
+int pwrseq_power_off(struct pwrseq_desc *desc)
+{
+	struct pwrseq_device *pwrseq;
+	struct pwrseq_unit *unit;
+	int ret;
+
+	might_sleep();
+
+	if (!desc || !desc->powered_on)
+		return 0;
+
+	pwrseq = desc->pwrseq;
+	unit = desc->target->unit;
+
+	guard(rwsem_read)(&pwrseq->rw_lock);
+	if (!device_is_registered(&pwrseq->dev))
+		return -ENODEV;
+
+	guard(mutex)(&pwrseq->state_lock);
+
+	ret = pwrseq_unit_disable(pwrseq, unit);
+	if (!ret)
+		desc->powered_on = false;
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(pwrseq_power_off);
+
+#if IS_ENABLED(CONFIG_DEBUG_FS)
+
+struct pwrseq_debugfs_count_ctx {
+	struct device *dev;
+	loff_t index;
+};
+
+static int pwrseq_debugfs_seq_count(struct device *dev, void *data)
+{
+	struct pwrseq_debugfs_count_ctx *ctx = data;
+
+	ctx->dev = dev;
+
+	return ctx->index-- ? 0 : 1;
+}
+
+static void *pwrseq_debugfs_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct pwrseq_debugfs_count_ctx ctx;
+
+	ctx.dev = NULL;
+	ctx.index = *pos;
+
+	/*
+	 * We're holding the lock for the entire printout so no need to fiddle
+	 * with device reference count.
+	 */
+	down_read(&pwrseq_sem);
+
+	bus_for_each_dev(&pwrseq_bus, NULL, &ctx, pwrseq_debugfs_seq_count);
+	if (!ctx.index)
+		return NULL;
+
+	return ctx.dev;
+}
+
+static void *pwrseq_debugfs_seq_next(struct seq_file *seq, void *data,
+				     loff_t *pos)
+{
+	struct device *curr = data;
+
+	++*pos;
+
+	struct device *next __free(put_device) =
+			bus_find_next_device(&pwrseq_bus, curr);
+	return next;
+}
+
+static void pwrseq_debugfs_seq_show_target(struct seq_file *seq,
+					   struct pwrseq_target *target)
+{
+	seq_printf(seq, "    target: [%s] (target unit: [%s])\n",
+		   target->name, target->unit->name);
+}
+
+static void pwrseq_debugfs_seq_show_unit(struct seq_file *seq,
+					 struct pwrseq_unit *unit)
+{
+	struct pwrseq_unit_dep *ref;
+
+	seq_printf(seq, "    unit: [%s] - enable count: %u\n",
+		   unit->name, unit->enable_count);
+
+	if (list_empty(&unit->deps))
+		return;
+
+	seq_puts(seq, "      dependencies:\n");
+	list_for_each_entry(ref, &unit->deps, list)
+		seq_printf(seq, "        [%s]\n", ref->unit->name);
+}
+
+static int pwrseq_debugfs_seq_show(struct seq_file *seq, void *data)
+{
+	struct device *dev = data;
+	struct pwrseq_device *pwrseq = to_pwrseq_device(dev);
+	struct pwrseq_target *target;
+	struct pwrseq_unit *unit;
+
+	seq_printf(seq, "%s:\n", dev_name(dev));
+
+	seq_puts(seq, "  targets:\n");
+	list_for_each_entry(target, &pwrseq->targets, list)
+		pwrseq_debugfs_seq_show_target(seq, target);
+
+	seq_puts(seq, "  units:\n");
+	list_for_each_entry(unit, &pwrseq->units, list)
+		pwrseq_debugfs_seq_show_unit(seq, unit);
+
+	return 0;
+}
+
+static void pwrseq_debugfs_seq_stop(struct seq_file *seq, void *data)
+{
+	up_read(&pwrseq_sem);
+}
+
+static const struct seq_operations pwrseq_debugfs_sops = {
+	.start = pwrseq_debugfs_seq_start,
+	.next = pwrseq_debugfs_seq_next,
+	.show = pwrseq_debugfs_seq_show,
+	.stop = pwrseq_debugfs_seq_stop,
+};
+DEFINE_SEQ_ATTRIBUTE(pwrseq_debugfs);
+
+static struct dentry *pwrseq_debugfs_dentry;
+
+#endif /* CONFIG_DEBUG_FS */
+
+static int __init pwrseq_init(void)
+{
+	int ret;
+
+	ret = bus_register(&pwrseq_bus);
+	if (ret) {
+		pr_err("Failed to register the power sequencer bus\n");
+		return ret;
+	}
+
+#if IS_ENABLED(CONFIG_DEBUG_FS)
+	pwrseq_debugfs_dentry = debugfs_create_file("pwrseq", 0444, NULL, NULL,
+						    &pwrseq_debugfs_fops);
+#endif  /* CONFIG_DEBUG_FS */
+
+	return 0;
+}
+subsys_initcall(pwrseq_init);
+
+static void __exit pwrseq_exit(void)
+{
+#if IS_ENABLED(CONFIG_DEBUG_FS)
+	debugfs_remove_recursive(pwrseq_debugfs_dentry);
+#endif  /* CONFIG_DEBUG_FS */
+
+	bus_unregister(&pwrseq_bus);
+}
+module_exit(pwrseq_exit);
+
+MODULE_AUTHOR("Bartosz Golaszewski <bartosz.golaszewski@linaro.org>");
+MODULE_DESCRIPTION("Power Sequencing subsystem core");
+MODULE_LICENSE("GPL");
diff --git a/include/linux/pwrseq/consumer.h b/include/linux/pwrseq/consumer.h
new file mode 100644
index 000000000000..7d583b4f266e
--- /dev/null
+++ b/include/linux/pwrseq/consumer.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2024 Linaro Ltd.
+ */
+
+#ifndef __POWER_SEQUENCING_CONSUMER_H__
+#define __POWER_SEQUENCING_CONSUMER_H__
+
+#include <linux/err.h>
+
+struct device;
+struct pwrseq_desc;
+
+#if IS_ENABLED(CONFIG_POWER_SEQUENCING)
+
+struct pwrseq_desc * __must_check
+pwrseq_get(struct device *dev, const char *target);
+void pwrseq_put(struct pwrseq_desc *desc);
+
+struct pwrseq_desc * __must_check
+devm_pwrseq_get(struct device *dev, const char *target);
+
+int pwrseq_power_on(struct pwrseq_desc *desc);
+int pwrseq_power_off(struct pwrseq_desc *desc);
+
+#else /* CONFIG_POWER_SEQUENCING */
+
+static inline struct pwrseq_desc * __must_check
+pwrseq_get(struct device *dev, const char *target)
+{
+	return ERR_PTR(-ENOSYS);
+}
+
+static inline void pwrseq_put(struct pwrseq_desc *desc)
+{
+}
+
+static inline struct pwrseq_desc * __must_check
+devm_pwrseq_get(struct device *dev, const char *target)
+{
+	return ERR_PTR(-ENOSYS);
+}
+
+static inline int pwrseq_power_on(struct pwrseq_desc *desc)
+{
+	return -ENOSYS;
+}
+
+static inline int pwrseq_power_off(struct pwrseq_desc *desc)
+{
+	return -ENOSYS;
+}
+
+#endif /* CONFIG_POWER_SEQUENCING */
+
+#endif /* __POWER_SEQUENCING_CONSUMER_H__ */
diff --git a/include/linux/pwrseq/provider.h b/include/linux/pwrseq/provider.h
new file mode 100644
index 000000000000..cbc3607cbfcf
--- /dev/null
+++ b/include/linux/pwrseq/provider.h
@@ -0,0 +1,75 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2024 Linaro Ltd.
+ */
+
+#ifndef __POWER_SEQUENCING_PROVIDER_H__
+#define __POWER_SEQUENCING_PROVIDER_H__
+
+struct device;
+struct module;
+struct pwrseq_device;
+
+typedef int (*pwrseq_power_state_func)(struct pwrseq_device *);
+typedef int (*pwrseq_match_func)(struct pwrseq_device *, struct device *);
+
+/**
+ * struct pwrseq_unit_data - Configuration of a single power sequencing
+ *                           unit.
+ * @name: Name of the unit.
+ * @deps: Units that must be enabled before this one and disabled after it
+ *        in the order they come in this array. Must be NULL-terminated.
+ * @enable: Callback running the part of the power-on sequence provided by
+ *          this unit.
+ * @disable: Callback running the part of the power-off sequence provided
+ *           by this unit.
+ */
+struct pwrseq_unit_data {
+	const char *name;
+	const struct pwrseq_unit_data **deps;
+	pwrseq_power_state_func enable;
+	pwrseq_power_state_func disable;
+};
+
+/**
+ * struct pwrseq_target_data - Configuration of a power sequencing target.
+ * @name: Name of the target.
+ * @unit: Final unit that this target must reach in order to be considered
+ *        enabled.
+ * @post_enable: Callback run after the target unit has been enabled, *after*
+ *               the state lock has been released. It's useful for implementing
+ *               boot-up delays without blocking other users from powering up
+ *               using the same power sequencer.
+ */
+struct pwrseq_target_data {
+	const char *name;
+	const struct pwrseq_unit_data *unit;
+	pwrseq_power_state_func post_enable;
+};
+
+/**
+ * struct pwrseq_config - Configuration used for registering a new provider.
+ * @parent: Parent device for the sequencer. Must be set.
+ * @owner: Module providing this device.
+ * @drvdata: Private driver data.
+ * @match: Provider callback used to match the consumer device to the sequencer.
+ * @targets: Array of targets for this power sequencer. Must be NULL-terminated.
+ */
+struct pwrseq_config {
+	struct device *parent;
+	struct module *owner;
+	void *drvdata;
+	pwrseq_match_func match;
+	const struct pwrseq_target_data **targets;
+};
+
+struct pwrseq_device *
+pwrseq_device_register(const struct pwrseq_config *config);
+void pwrseq_device_unregister(struct pwrseq_device *pwrseq);
+struct pwrseq_device *
+devm_pwrseq_device_register(struct device *dev,
+			    const struct pwrseq_config *config);
+
+void *pwrseq_device_get_drvdata(struct pwrseq_device *pwrseq);
+
+#endif /* __POWER_SEQUENCING_PROVIDER_H__ */
-- 
2.40.1

[PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

This adds the power sequencing driver for the PMU modules present on the
Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
and knows how to match the sequencer to the consumer device by verifying
the relevant properties and DT layout. Using this driver will allow the
BT and WLAN drivers to respect the required delays between enabling the
two modules.

Tested-by: Amit Pundir <amit.pundir@linaro.org>
Tested-by: Neil Armstrong <neil.armstrong@linaro.org> # on SM8550-QRD, SM8650-QRD & SM8650-HDK
Tested-by: Caleb Connolly <caleb.connolly@linaro.org> # OnePlus 8T
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 drivers/power/sequencing/Kconfig           |  17 ++
 drivers/power/sequencing/Makefile          |   2 +
 drivers/power/sequencing/pwrseq-qcom-wcn.c | 336 +++++++++++++++++++++
 3 files changed, 355 insertions(+)
 create mode 100644 drivers/power/sequencing/pwrseq-qcom-wcn.c

diff --git a/drivers/power/sequencing/Kconfig b/drivers/power/sequencing/Kconfig
index ba5732b1dbf8..c9f1cdb66524 100644
--- a/drivers/power/sequencing/Kconfig
+++ b/drivers/power/sequencing/Kconfig
@@ -10,3 +10,20 @@ menuconfig POWER_SEQUENCING
 	  during power-up.
 
 	  If unsure, say no.
+
+if POWER_SEQUENCING
+
+config POWER_SEQUENCING_QCOM_WCN
+	tristate "Qualcomm WCN family PMU driver"
+	default m if ARCH_QCOM
+	help
+	  Say Y here to enable the power sequencing driver for Qualcomm
+	  WCN Bluetooth/WLAN chipsets.
+
+	  Typically, a package from the Qualcomm WCN family contains the BT
+	  and WLAN modules whose power is controlled by the PMU module. As the
+	  former two share the power-up sequence which is executed by the PMU,
+	  this driver is needed for correct power control or else we'd risk not
+	  respecting the required delays between enabling Bluetooth and WLAN.
+
+endif
diff --git a/drivers/power/sequencing/Makefile b/drivers/power/sequencing/Makefile
index dcdf8c0c159e..2eec2df7912d 100644
--- a/drivers/power/sequencing/Makefile
+++ b/drivers/power/sequencing/Makefile
@@ -2,3 +2,5 @@
 
 obj-$(CONFIG_POWER_SEQUENCING)		+= pwrseq-core.o
 pwrseq-core-y				:= core.o
+
+obj-$(CONFIG_POWER_SEQUENCING_QCOM_WCN)	+= pwrseq-qcom-wcn.o
diff --git a/drivers/power/sequencing/pwrseq-qcom-wcn.c b/drivers/power/sequencing/pwrseq-qcom-wcn.c
new file mode 100644
index 000000000000..42dacfda745e
--- /dev/null
+++ b/drivers/power/sequencing/pwrseq-qcom-wcn.c
@@ -0,0 +1,336 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2024 Linaro Ltd.
+ */
+
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/gpio/consumer.h>
+#include <linux/jiffies.h>
+#include <linux/mod_devicetable.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/regulator/consumer.h>
+#include <linux/pwrseq/provider.h>
+#include <linux/string.h>
+#include <linux/types.h>
+
+struct pwrseq_qcom_wcn_pdata {
+	const char *const *vregs;
+	size_t num_vregs;
+	unsigned int pwup_delay_ms;
+	unsigned int gpio_enable_delay_ms;
+};
+
+struct pwrseq_qcom_wcn_ctx {
+	struct pwrseq_device *pwrseq;
+	struct device_node *of_node;
+	const struct pwrseq_qcom_wcn_pdata *pdata;
+	struct regulator_bulk_data *regs;
+	struct gpio_desc *bt_gpio;
+	struct gpio_desc *wlan_gpio;
+	struct clk *clk;
+	unsigned long last_gpio_enable_jf;
+};
+
+static void pwrseq_qcom_wcn_ensure_gpio_delay(struct pwrseq_qcom_wcn_ctx *ctx)
+{
+	unsigned long diff_jiffies;
+	unsigned int diff_msecs;
+
+	if (!ctx->pdata->gpio_enable_delay_ms)
+		return;
+
+	diff_jiffies = jiffies - ctx->last_gpio_enable_jf;
+	diff_msecs = jiffies_to_msecs(diff_jiffies);
+
+	if (diff_msecs < ctx->pdata->gpio_enable_delay_ms)
+		msleep(ctx->pdata->gpio_enable_delay_ms - diff_msecs);
+}
+
+static int pwrseq_qcom_wcn_vregs_enable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	return regulator_bulk_enable(ctx->pdata->num_vregs, ctx->regs);
+}
+
+static int pwrseq_qcom_wcn_vregs_disable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	return regulator_bulk_disable(ctx->pdata->num_vregs, ctx->regs);
+}
+
+static const struct pwrseq_unit_data pwrseq_qcom_wcn_vregs_unit_data = {
+	.name = "regulators-enable",
+	.enable = pwrseq_qcom_wcn_vregs_enable,
+	.disable = pwrseq_qcom_wcn_vregs_disable,
+};
+
+static int pwrseq_qcom_wcn_clk_enable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	return clk_prepare_enable(ctx->clk);
+}
+
+static int pwrseq_qcom_wcn_clk_disable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	clk_disable_unprepare(ctx->clk);
+
+	return 0;
+}
+
+static const struct pwrseq_unit_data pwrseq_qcom_wcn_clk_unit_data = {
+	.name = "clock-enable",
+	.enable = pwrseq_qcom_wcn_clk_enable,
+	.disable = pwrseq_qcom_wcn_clk_disable,
+};
+
+static const struct pwrseq_unit_data *pwrseq_qcom_wcn_unit_deps[] = {
+	&pwrseq_qcom_wcn_vregs_unit_data,
+	&pwrseq_qcom_wcn_clk_unit_data,
+	NULL
+};
+
+static int pwrseq_qcom_wcn_bt_enable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	pwrseq_qcom_wcn_ensure_gpio_delay(ctx);
+	gpiod_set_value_cansleep(ctx->bt_gpio, 1);
+	ctx->last_gpio_enable_jf = jiffies;
+
+	return 0;
+}
+
+static int pwrseq_qcom_wcn_bt_disable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	gpiod_set_value_cansleep(ctx->bt_gpio, 0);
+
+	return 0;
+}
+
+static const struct pwrseq_unit_data pwrseq_qcom_wcn_bt_unit_data = {
+	.name = "bluetooth-enable",
+	.deps = pwrseq_qcom_wcn_unit_deps,
+	.enable = pwrseq_qcom_wcn_bt_enable,
+	.disable = pwrseq_qcom_wcn_bt_disable,
+};
+
+static int pwrseq_qcom_wcn_wlan_enable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	pwrseq_qcom_wcn_ensure_gpio_delay(ctx);
+	gpiod_set_value_cansleep(ctx->wlan_gpio, 1);
+	ctx->last_gpio_enable_jf = jiffies;
+
+	return 0;
+}
+
+static int pwrseq_qcom_wcn_wlan_disable(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	gpiod_set_value_cansleep(ctx->wlan_gpio, 0);
+
+	return 0;
+}
+
+static const struct pwrseq_unit_data pwrseq_qcom_wcn_wlan_unit_data = {
+	.name = "wlan-enable",
+	.deps = pwrseq_qcom_wcn_unit_deps,
+	.enable = pwrseq_qcom_wcn_wlan_enable,
+	.disable = pwrseq_qcom_wcn_wlan_disable,
+};
+
+static int pwrseq_qcom_wcn_pwup_delay(struct pwrseq_device *pwrseq)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+
+	if (ctx->pdata->pwup_delay_ms)
+		msleep(ctx->pdata->pwup_delay_ms);
+
+	return 0;
+}
+
+static const struct pwrseq_target_data pwrseq_qcom_wcn_bt_target_data = {
+	.name = "bluetooth",
+	.unit = &pwrseq_qcom_wcn_bt_unit_data,
+	.post_enable = pwrseq_qcom_wcn_pwup_delay,
+};
+
+static const struct pwrseq_target_data pwrseq_qcom_wcn_wlan_target_data = {
+	.name = "wlan",
+	.unit = &pwrseq_qcom_wcn_wlan_unit_data,
+	.post_enable = pwrseq_qcom_wcn_pwup_delay,
+};
+
+static const struct pwrseq_target_data *pwrseq_qcom_wcn_targets[] = {
+	&pwrseq_qcom_wcn_bt_target_data,
+	&pwrseq_qcom_wcn_wlan_target_data,
+	NULL
+};
+
+static const char *const pwrseq_qca6390_vregs[] = {
+	"vddio",
+	"vddaon",
+	"vddpmu",
+	"vddrfa0p95",
+	"vddrfa1p3",
+	"vddrfa1p9",
+	"vddpcie1p3",
+	"vddpcie1p9",
+};
+
+static const struct pwrseq_qcom_wcn_pdata pwrseq_qca6390_of_data = {
+	.vregs = pwrseq_qca6390_vregs,
+	.num_vregs = ARRAY_SIZE(pwrseq_qca6390_vregs),
+	.pwup_delay_ms = 60,
+	.gpio_enable_delay_ms = 100,
+};
+
+static const char *const pwrseq_wcn7850_vregs[] = {
+	"vdd",
+	"vddio",
+	"vddio1p2",
+	"vddaon",
+	"vdddig",
+	"vddrfa1p2",
+	"vddrfa1p8",
+};
+
+static const struct pwrseq_qcom_wcn_pdata pwrseq_wcn7850_of_data = {
+	.vregs = pwrseq_wcn7850_vregs,
+	.num_vregs = ARRAY_SIZE(pwrseq_wcn7850_vregs),
+	.pwup_delay_ms = 50,
+};
+
+static int pwrseq_qcom_wcn_match(struct pwrseq_device *pwrseq,
+				 struct device *dev)
+{
+	struct pwrseq_qcom_wcn_ctx *ctx = pwrseq_device_get_drvdata(pwrseq);
+	struct device_node *dev_node = dev->of_node;
+
+	/*
+	 * The PMU supplies power to the Bluetooth and WLAN modules. both
+	 * consume the PMU AON output so check the presence of the
+	 * 'vddaon-supply' property and whether it leads us to the right
+	 * device.
+	 */
+	if (!of_property_present(dev_node, "vddaon-supply"))
+		return 0;
+
+	struct device_node *reg_node __free(device_node) =
+			of_parse_phandle(dev_node, "vddaon-supply", 0);
+	if (!reg_node)
+		return 0;
+
+	/*
+	 * `reg_node` is the PMU AON regulator, its parent is the `regulators`
+	 * node and finally its grandparent is the PMU device node that we're
+	 * looking for.
+	 */
+	if (!reg_node->parent || !reg_node->parent->parent ||
+	    reg_node->parent->parent != ctx->of_node)
+		return 0;
+
+	return 1;
+}
+
+static int pwrseq_qcom_wcn_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct pwrseq_qcom_wcn_ctx *ctx;
+	struct pwrseq_config config;
+	int i, ret;
+
+	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	ctx->of_node = dev->of_node;
+
+	ctx->pdata = of_device_get_match_data(dev);
+	if (!ctx->pdata)
+		return dev_err_probe(dev, -ENODEV,
+				     "Failed to obtain platform data\n");
+
+	ctx->regs = devm_kcalloc(dev, ctx->pdata->num_vregs,
+				 sizeof(*ctx->regs), GFP_KERNEL);
+	if (!ctx->regs)
+		return -ENOMEM;
+
+	for (i = 0; i < ctx->pdata->num_vregs; i++)
+		ctx->regs[i].supply = ctx->pdata->vregs[i];
+
+	ret = devm_regulator_bulk_get(dev, ctx->pdata->num_vregs, ctx->regs);
+	if (ret < 0)
+		return dev_err_probe(dev, ret,
+				     "Failed to get all regulators\n");
+
+	ctx->bt_gpio = devm_gpiod_get_optional(dev, "bt-enable", GPIOD_OUT_LOW);
+	if (IS_ERR(ctx->bt_gpio))
+		return dev_err_probe(dev, PTR_ERR(ctx->bt_gpio),
+				     "Failed to get the Bluetooth enable GPIO\n");
+
+	ctx->wlan_gpio = devm_gpiod_get_optional(dev, "wlan-enable",
+						 GPIOD_OUT_LOW);
+	if (IS_ERR(ctx->wlan_gpio))
+		return dev_err_probe(dev, PTR_ERR(ctx->wlan_gpio),
+				     "Failed to get the WLAN enable GPIO\n");
+
+	ctx->clk = devm_clk_get_optional(dev, NULL);
+	if (IS_ERR(ctx->clk))
+		return dev_err_probe(dev, PTR_ERR(ctx->clk),
+				     "Failed to get the reference clock\n");
+
+	memset(&config, 0, sizeof(config));
+
+	config.parent = dev;
+	config.owner = THIS_MODULE;
+	config.drvdata = ctx;
+	config.match = pwrseq_qcom_wcn_match;
+	config.targets = pwrseq_qcom_wcn_targets;
+
+	ctx->pwrseq = devm_pwrseq_device_register(dev, &config);
+	if (IS_ERR(ctx->pwrseq))
+		return dev_err_probe(dev, PTR_ERR(ctx->pwrseq),
+				     "Failed to register the power sequencer\n");
+
+	return 0;
+}
+
+static const struct of_device_id pwrseq_qcom_wcn_of_match[] = {
+	{
+		.compatible = "qcom,qca6390-pmu",
+		.data = &pwrseq_qca6390_of_data,
+	},
+	{
+		.compatible = "qcom,wcn7850-pmu",
+		.data = &pwrseq_wcn7850_of_data,
+	},
+	{ }
+};
+MODULE_DEVICE_TABLE(of, pwrseq_qcom_wcn_of_match);
+
+static struct platform_driver pwrseq_qcom_wcn_driver = {
+	.driver = {
+		.name = "pwrseq-qcom_wcn",
+		.of_match_table = pwrseq_qcom_wcn_of_match,
+	},
+	.probe = pwrseq_qcom_wcn_probe,
+};
+module_platform_driver(pwrseq_qcom_wcn_driver);
+
+MODULE_AUTHOR("Bartosz Golaszewski <bartosz.golaszewski@linaro.org>");
+MODULE_DESCRIPTION("Qualcomm WCN PMU power sequencing driver");
+MODULE_LICENSE("GPL");
-- 
2.40.1

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Bartosz Golaszewski]

On Wed, Jun 5, 2024 at 2:39 PM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
>
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>
> Hi!
>
> These are the power sequencing patches sent separately after some
> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
> new branch and maintain the subsystem from now on. I then plan to
> provide an immutable tag to the Bluetooth and PCI subsystems so that the
> rest of the C changes can be applied. This new branch will then be
> directly sent to Linus Torvalds for the next merge window.
>
> Changelog:
>
> Since v8:
> - split the pwrseq patches out into their own series
> - rename incref/decref functions to get/put for consistency
> - fix typos
> - make it very explicit in docs that arrays of targets and units must be
>   NULL-terminated
> - Link to v8: https://lore.kernel.org/r/20240528-pwrseq-v8-0-d354d52b763c@linaro.org
>
> Since v7:
> - added DTS changes for sm8650-hdk
> - added circular dependency detection for pwrseq units
> - fixed a KASAN reported use-after-free error in remove path
> - improve Kconfig descriptions
> - fix typos in bindings and Kconfig
> - fixed issues reported by smatch
> - fix the unbind path in PCI pwrctl
> - lots of minor improvements to the pwrseq core
>
> Since v6:
> - kernel doc fixes
> - drop myself from the DT bindings maintainers list for ath12k
> - wait until the PCI bridge device is fully added before creating the
>   PCI pwrctl platform devices for its sub-nodes, otherwise we may see
>   sysfs and procfs attribute failures (due to duplication, we're
>   basically trying to probe the same device twice at the same time)
> - I kept the regulators for QCA6390's ath11k as required as they only
>   apply to this specific Qualcomm package
>
> Since v5:
> - unify the approach to modelling the WCN WLAN/BT chips by always exposing
>   the PMU node on the device tree and making the WLAN and BT nodes become
>   consumers of its power outputs; this includes a major rework of the DT
>   sources, bindings and driver code; there's no more a separate PCI
>   pwrctl driver for WCN7850, instead its power-up sequence was moved
>   into the pwrseq driver common for all WCN chips
> - don't set load_uA from new regulator consumers
> - fix reported kerneldoc issues
> - drop voltage ranges for PMU outputs from DT
> - many minor tweaks and reworks
>
> v1: Original RFC:
>
> https://lore.kernel.org/lkml/20240104130123.37115-1-brgl@bgdev.pl/T/
>
> v2: First real patch series (should have been PATCH v2) adding what I
>     referred to back then as PCI power sequencing:
>
> https://lore.kernel.org/linux-arm-kernel/2024021413-grumbling-unlivable-c145@gregkh/T/
>
> v3: RFC for the DT representation of the PMU supplying the WLAN and BT
>     modules inside the QCA6391 package (was largely separate from the
>     series but probably should have been called PATCH or RFC v3):
>
> https://lore.kernel.org/all/CAMRc=Mc+GNoi57eTQg71DXkQKjdaoAmCpB=h2ndEpGnmdhVV-Q@mail.gmail.com/T/
>
> v4: Second attempt at the full series with changed scope (introduction of
>     the pwrseq subsystem, should have been RFC v4)
>
> https://lore.kernel.org/lkml/20240201155532.49707-1-brgl@bgdev.pl/T/
>
> v5: Two different ways of handling QCA6390 and WCN7850:
>
> https://lore.kernel.org/lkml/20240216203215.40870-1-brgl@bgdev.pl/
>
> Bartosz Golaszewski (2):
>   power: sequencing: implement the pwrseq core
>   power: pwrseq: add a driver for the PMU module on the QCom WCN
>     chipsets
>
>  MAINTAINERS                                |    8 +
>  drivers/power/Kconfig                      |    1 +
>  drivers/power/Makefile                     |    1 +
>  drivers/power/sequencing/Kconfig           |   29 +
>  drivers/power/sequencing/Makefile          |    6 +
>  drivers/power/sequencing/core.c            | 1105 ++++++++++++++++++++
>  drivers/power/sequencing/pwrseq-qcom-wcn.c |  336 ++++++
>  include/linux/pwrseq/consumer.h            |   56 +
>  include/linux/pwrseq/provider.h            |   75 ++
>  9 files changed, 1617 insertions(+)
>  create mode 100644 drivers/power/sequencing/Kconfig
>  create mode 100644 drivers/power/sequencing/Makefile
>  create mode 100644 drivers/power/sequencing/core.c
>  create mode 100644 drivers/power/sequencing/pwrseq-qcom-wcn.c
>  create mode 100644 include/linux/pwrseq/consumer.h
>  create mode 100644 include/linux/pwrseq/provider.h
>
> --
> 2.40.1
>

With no objections, I plan to pick it up into its own pwrseq branch on
Wednesday.

Review and Ack tags are welcome.

Bartosz

Re: [PATCH v9 1/2] power: sequencing: implement the pwrseq core

[Krzysztof Kozlowski]

On 05/06/2024 14:38, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Implement the power sequencing subsystem allowing devices to share
> complex powering-up and down procedures. It's split into the consumer
> and provider parts but does not implement any new DT bindings so that
> the actual power sequencing is never revealed in the DT representation.
> 
> Tested-by: Amit Pundir <amit.pundir@linaro.org>
> Tested-by: Neil Armstrong <neil.armstrong@linaro.org> # on SM8550-QRD, SM8650-QRD & SM8650-HDK
> Tested-by: Caleb Connolly <caleb.connolly@linaro.org> # OnePlus 8T
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Looks fine to me, including the concept and solving real problems.

Acked-by: Krzysztof Kozlowski <krzk@kernel.org>

Best regards,
Krzysztof

Re: [PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Krzysztof Kozlowski]

On 05/06/2024 14:38, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> This adds the power sequencing driver for the PMU modules present on the
> Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
> and knows how to match the sequencer to the consumer device by verifying
> the relevant properties and DT layout. Using this driver will allow the
> BT and WLAN drivers to respect the required delays between enabling the
> two modules.

...

> +
> +static const struct of_device_id pwrseq_qcom_wcn_of_match[] = {
> +	{
> +		.compatible = "qcom,qca6390-pmu",
> +		.data = &pwrseq_qca6390_of_data,

Bindings *must* be part of this patchset. I missed the moment this was
split into such chunks.

Best regards,
Krzysztof

Re: [PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Bartosz Golaszewski]

On Mon, Jun 10, 2024 at 4:54 PM Krzysztof Kozlowski
<krzysztof.kozlowski@linaro.org> wrote:
>
> On 05/06/2024 14:38, Bartosz Golaszewski wrote:
> > From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> >
> > This adds the power sequencing driver for the PMU modules present on the
> > Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
> > and knows how to match the sequencer to the consumer device by verifying
> > the relevant properties and DT layout. Using this driver will allow the
> > BT and WLAN drivers to respect the required delays between enabling the
> > two modules.
>
> ...
>
> > +
> > +static const struct of_device_id pwrseq_qcom_wcn_of_match[] = {
> > +     {
> > +             .compatible = "qcom,qca6390-pmu",
> > +             .data = &pwrseq_qca6390_of_data,
>
> Bindings *must* be part of this patchset. I missed the moment this was
> split into such chunks.
>
> Best regards,
> Krzysztof
>

The bindings are already in next via Mark Brown's tree.

Bart

Re: [PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Krzysztof Kozlowski]

On 10/06/2024 16:55, Bartosz Golaszewski wrote:
> On Mon, Jun 10, 2024 at 4:54 PM Krzysztof Kozlowski
> <krzysztof.kozlowski@linaro.org> wrote:
>>
>> On 05/06/2024 14:38, Bartosz Golaszewski wrote:
>>> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>>
>>> This adds the power sequencing driver for the PMU modules present on the
>>> Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
>>> and knows how to match the sequencer to the consumer device by verifying
>>> the relevant properties and DT layout. Using this driver will allow the
>>> BT and WLAN drivers to respect the required delays between enabling the
>>> two modules.
>>
>> ...
>>
>>> +
>>> +static const struct of_device_id pwrseq_qcom_wcn_of_match[] = {
>>> +     {
>>> +             .compatible = "qcom,qca6390-pmu",
>>> +             .data = &pwrseq_qca6390_of_data,
>>
>> Bindings *must* be part of this patchset. I missed the moment this was
>> split into such chunks.
>>
>> Best regards,
>> Krzysztof
>>
> 
> The bindings are already in next via Mark Brown's tree.

OK, I checked linux-next from a day before Mark merged it. Cover letter
should mention it, because it is a bit unusual (and would save you
reviewer's question).

Best regards,
Krzysztof

Re: [PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Krzysztof Kozlowski]

On 05/06/2024 14:38, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> This adds the power sequencing driver for the PMU modules present on the
> Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
> and knows how to match the sequencer to the consumer device by verifying
> the relevant properties and DT layout. Using this driver will allow the
> BT and WLAN drivers to respect the required delays between enabling the
> two modules.
> 

...


> +
> +	ctx->bt_gpio = devm_gpiod_get_optional(dev, "bt-enable", GPIOD_OUT_LOW);
> +	if (IS_ERR(ctx->bt_gpio))
> +		return dev_err_probe(dev, PTR_ERR(ctx->bt_gpio),
> +				     "Failed to get the Bluetooth enable GPIO\n");
> +
> +	ctx->wlan_gpio = devm_gpiod_get_optional(dev, "wlan-enable",
> +						 GPIOD_OUT_LOW);
> +	if (IS_ERR(ctx->wlan_gpio))
> +		return dev_err_probe(dev, PTR_ERR(ctx->wlan_gpio),
> +				     "Failed to get the WLAN enable GPIO\n");
> +
> +	ctx->clk = devm_clk_get_optional(dev, NULL);

Your binding does not allow the clock. Do you need it for non-DT platforms?

Best regards,
Krzysztof

Re: [PATCH v9 2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets

[Krzysztof Kozlowski]

On 10/06/2024 17:05, Krzysztof Kozlowski wrote:
> On 05/06/2024 14:38, Bartosz Golaszewski wrote:
>> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>
>> This adds the power sequencing driver for the PMU modules present on the
>> Qualcomm WCN Bluetooth and Wifi chipsets. It uses the pwrseq subsystem
>> and knows how to match the sequencer to the consumer device by verifying
>> the relevant properties and DT layout. Using this driver will allow the
>> BT and WLAN drivers to respect the required delays between enabling the
>> two modules.
>>
> 
> ...
> 
> 
>> +
>> +	ctx->bt_gpio = devm_gpiod_get_optional(dev, "bt-enable", GPIOD_OUT_LOW);
>> +	if (IS_ERR(ctx->bt_gpio))
>> +		return dev_err_probe(dev, PTR_ERR(ctx->bt_gpio),
>> +				     "Failed to get the Bluetooth enable GPIO\n");
>> +
>> +	ctx->wlan_gpio = devm_gpiod_get_optional(dev, "wlan-enable",
>> +						 GPIOD_OUT_LOW);
>> +	if (IS_ERR(ctx->wlan_gpio))
>> +		return dev_err_probe(dev, PTR_ERR(ctx->wlan_gpio),
>> +				     "Failed to get the WLAN enable GPIO\n");
>> +
>> +	ctx->clk = devm_clk_get_optional(dev, NULL);
> 
> Your binding does not allow the clock. Do you need it for non-DT platforms?

Bah, damn covid, now I see second commit with clocks.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>

Best regards,
Krzysztof

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>


On Wed, 05 Jun 2024 14:38:48 +0200, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Hi!
> 
> These are the power sequencing patches sent separately after some
> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
> new branch and maintain the subsystem from now on. I then plan to
> provide an immutable tag to the Bluetooth and PCI subsystems so that the
> rest of the C changes can be applied. This new branch will then be
> directly sent to Linus Torvalds for the next merge window.
> 
> [...]

Applied, thanks!

[1/2] power: sequencing: implement the pwrseq core
      commit: 249ebf3f65f8530beb2cbfb91bff1d83ba88d23c
[2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
      commit: 2f1630f437dff20d02e4b3f07e836f42869128dd

Best regards,
-- 
Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[patchwork-bot+bluetooth]

Hello:

This series was applied to bluetooth/bluetooth-next.git (master)
by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:

On Wed,  5 Jun 2024 14:38:48 +0200 you wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Hi!
> 
> These are the power sequencing patches sent separately after some
> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
> new branch and maintain the subsystem from now on. I then plan to
> provide an immutable tag to the Bluetooth and PCI subsystems so that the
> rest of the C changes can be applied. This new branch will then be
> directly sent to Linus Torvalds for the next merge window.
> 
> [...]

Here is the summary with links:
  - [v9,1/2] power: sequencing: implement the pwrseq core
    https://git.kernel.org/bluetooth/bluetooth-next/c/249ebf3f65f8
  - [v9,2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
    https://git.kernel.org/bluetooth/bluetooth-next/c/2f1630f437df

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Lk Sii]

On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
> Hello:
> 
> This series was applied to bluetooth/bluetooth-next.git (master)
> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
> 
Hi luiz,

i am curious why Bartosz is able to merge his changes into bluetooth
development tree bluetooth-next directly.

1)
his changes should belong to *POWER* scope instead of *Bluetooth*
obviously, however, there are *NOT* any SOB tag from either power and
bluetooth maintainer. these changes currently only have below Acked-by
and Signed-off-by tags:

Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

2)
his changes have not merged into linus mainline tree yet.

3)
perhaps, it is safer to pull his changes from linus mainline tree when
merged than to merge into bluetooth-next firstly.

> On Wed,  5 Jun 2024 14:38:48 +0200 you wrote:
>> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>
>> Hi!
>>
>> These are the power sequencing patches sent separately after some
>> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
>> new branch and maintain the subsystem from now on. I then plan to
>> provide an immutable tag to the Bluetooth and PCI subsystems so that the
>> rest of the C changes can be applied. This new branch will then be
>> directly sent to Linus Torvalds for the next merge window.
>>
>> [...]
> 
> Here is the summary with links:
>   - [v9,1/2] power: sequencing: implement the pwrseq core
>     https://git.kernel.org/bluetooth/bluetooth-next/c/249ebf3f65f8
>   - [v9,2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
>     https://git.kernel.org/bluetooth/bluetooth-next/c/2f1630f437df
> 
> You are awesome, thank you!

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Bartosz Golaszewski]

On Fri, Jun 21, 2024 at 3:14 AM Lk Sii <lk_sii@163.com> wrote:
>
>
>
> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
> > Hello:
> >
> > This series was applied to bluetooth/bluetooth-next.git (master)
> > by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
> >
> Hi luiz,
>
> i am curious why Bartosz is able to merge his changes into bluetooth
> development tree bluetooth-next directly.
>

This conversation is getting progressively worse...

> 1)
> his changes should belong to *POWER* scope instead of *Bluetooth*
> obviously, however, there are *NOT* any SOB tag from either power and
> bluetooth maintainer. these changes currently only have below Acked-by
> and Signed-off-by tags:
>
> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>

It's a new subsystem that has been discussed and reviewed for months
and thoroughly tested. Please refer to the cover letter under v8
linked in this thread. It's not related to power-management or
power-supply, it's its own thing but IMO the best place to put it is
under drivers/power/. And I will maintain it.

> 2)
> his changes have not merged into linus mainline tree yet.
>

This is why they are in next! They are scheduled to go in during the
upcoming merge window. But since changes belong in multiple trees, we
need a cross-tree merge.

> 3)
> perhaps, it is safer to pull his changes from linus mainline tree when
> merged than to merge into bluetooth-next firstly.
>

It's not safer at all, why would spending less time in next be safer?

> > On Wed,  5 Jun 2024 14:38:48 +0200 you wrote:
> >> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> >>
> >> Hi!
> >>
> >> These are the power sequencing patches sent separately after some
> >> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
> >> new branch and maintain the subsystem from now on. I then plan to
> >> provide an immutable tag to the Bluetooth and PCI subsystems so that the
> >> rest of the C changes can be applied. This new branch will then be
> >> directly sent to Linus Torvalds for the next merge window.
> >>
> >> [...]
> >
> > Here is the summary with links:
> >   - [v9,1/2] power: sequencing: implement the pwrseq core
> >     https://git.kernel.org/bluetooth/bluetooth-next/c/249ebf3f65f8
> >   - [v9,2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
> >     https://git.kernel.org/bluetooth/bluetooth-next/c/2f1630f437df
> >
> > You are awesome, thank you!
>

Why are you top-posting anyway?

Bart

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Lk Sii]

On 2024/6/21 14:36, Bartosz Golaszewski wrote:
> On Fri, Jun 21, 2024 at 3:14 AM Lk Sii <lk_sii@163.com> wrote:
>>
>>
>>
>> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
>>> Hello:
>>>
>>> This series was applied to bluetooth/bluetooth-next.git (master)
>>> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
>>>
>> Hi luiz,
>>
>> i am curious why Bartosz is able to merge his changes into bluetooth
>> development tree bluetooth-next directly.
>>
> 
> This conversation is getting progressively worse...
> 
>> 1)
>> his changes should belong to *POWER* scope instead of *Bluetooth*
>> obviously, however, there are *NOT* any SOB tag from either power and
>> bluetooth maintainer. these changes currently only have below Acked-by
>> and Signed-off-by tags:
>>
>> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
>> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>
> 
> It's a new subsystem that has been discussed and reviewed for months
> and thoroughly tested. Please refer to the cover letter under v8
> linked in this thread. It's not related to power-management or
> power-supply, it's its own thing but IMO the best place to put it is
> under drivers/power/. And I will maintain it.
> 
>> 2)
>> his changes have not merged into linus mainline tree yet.
>>
> 
> This is why they are in next! They are scheduled to go in during the
> upcoming merge window. But since changes belong in multiple trees, we
> need a cross-tree merge.
> 
>> 3)
>> perhaps, it is safer to pull his changes from linus mainline tree when
>> merged than to merge into bluetooth-next firstly.
>>
> 
> It's not safer at all, why would spending less time in next be safer?
> 
it seems this patch serial(new subsystem) does not depend on bluetooth
and also does not belong to bluetooth subsystem, but have been contained
by tip of bluetooth tree.

why not follow below merging produce?
1) you send this patch serials to Linus to merge within linus mainline tree
2) luiz then pull your changes from linus mainline tree.

>>> On Wed,  5 Jun 2024 14:38:48 +0200 you wrote:
>>>> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>>>
>>>> Hi!
>>>>
>>>> These are the power sequencing patches sent separately after some
>>>> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
>>>> new branch and maintain the subsystem from now on. I then plan to
>>>> provide an immutable tag to the Bluetooth and PCI subsystems so that the
>>>> rest of the C changes can be applied. This new branch will then be
>>>> directly sent to Linus Torvalds for the next merge window.
>>>>
>>>> [...]
>>>
>>> Here is the summary with links:
>>>   - [v9,1/2] power: sequencing: implement the pwrseq core
>>>     https://git.kernel.org/bluetooth/bluetooth-next/c/249ebf3f65f8
>>>   - [v9,2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
>>>     https://git.kernel.org/bluetooth/bluetooth-next/c/2f1630f437df
>>>
>>> You are awesome, thank you!
>>
> 
> Why are you top-posting anyway?
> 
it is caused by my bad mail client settings. thanks for reminder.
> Bart

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Bartosz Golaszewski]

On Fri, Jun 21, 2024 at 11:04 AM Lk Sii <lk_sii@163.com> wrote:
>
> On 2024/6/21 14:36, Bartosz Golaszewski wrote:
> > On Fri, Jun 21, 2024 at 3:14 AM Lk Sii <lk_sii@163.com> wrote:
> >>
> >>
> >>
> >> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
> >>> Hello:
> >>>
> >>> This series was applied to bluetooth/bluetooth-next.git (master)
> >>> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
> >>>
> >> Hi luiz,
> >>
> >> i am curious why Bartosz is able to merge his changes into bluetooth
> >> development tree bluetooth-next directly.
> >>
> >
> > This conversation is getting progressively worse...
> >
> >> 1)
> >> his changes should belong to *POWER* scope instead of *Bluetooth*
> >> obviously, however, there are *NOT* any SOB tag from either power and
> >> bluetooth maintainer. these changes currently only have below Acked-by
> >> and Signed-off-by tags:
> >>
> >> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
> >> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> >>
> >
> > It's a new subsystem that has been discussed and reviewed for months
> > and thoroughly tested. Please refer to the cover letter under v8
> > linked in this thread. It's not related to power-management or
> > power-supply, it's its own thing but IMO the best place to put it is
> > under drivers/power/. And I will maintain it.
> >
> >> 2)
> >> his changes have not merged into linus mainline tree yet.
> >>
> >
> > This is why they are in next! They are scheduled to go in during the
> > upcoming merge window. But since changes belong in multiple trees, we
> > need a cross-tree merge.
> >
> >> 3)
> >> perhaps, it is safer to pull his changes from linus mainline tree when
> >> merged than to merge into bluetooth-next firstly.
> >>
> >
> > It's not safer at all, why would spending less time in next be safer?
> >
> it seems this patch serial(new subsystem) does not depend on bluetooth
> and also does not belong to bluetooth subsystem, but have been contained
> by tip of bluetooth tree.
>

It's the other way around: bluetooth changes (namely the hci_qca
driver) depend on the power sequencing changes.

> why not follow below merging produce?
> 1) you send this patch serials to Linus to merge within linus mainline tree
> 2) luiz then pull your changes from linus mainline tree.
>

I explained this in my previous email. Why would you want these
changes to needlessly wait for another release cycle? It makes no
sense. It's just a regular cross-tree merge like hundreds that are
performed every release.

> >>> On Wed,  5 Jun 2024 14:38:48 +0200 you wrote:
> >>>> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> >>>>
> >>>> Hi!
> >>>>
> >>>> These are the power sequencing patches sent separately after some
> >>>> improvements suggested by Bjorn Helgaas. I intend to pick them up into a
> >>>> new branch and maintain the subsystem from now on. I then plan to
> >>>> provide an immutable tag to the Bluetooth and PCI subsystems so that the
> >>>> rest of the C changes can be applied. This new branch will then be
> >>>> directly sent to Linus Torvalds for the next merge window.
> >>>>
> >>>> [...]
> >>>
> >>> Here is the summary with links:
> >>>   - [v9,1/2] power: sequencing: implement the pwrseq core
> >>>     https://git.kernel.org/bluetooth/bluetooth-next/c/249ebf3f65f8
> >>>   - [v9,2/2] power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets
> >>>     https://git.kernel.org/bluetooth/bluetooth-next/c/2f1630f437df
> >>>
> >>> You are awesome, thank you!
> >>
> >
> > Why are you top-posting anyway?
> >
> it is caused by my bad mail client settings. thanks for reminder.
> > Bart
>

Luiz, Marcel: Am I wasting my time with this person? Is this another
Markus Elfring and I unknowingly got pulled into a nonsensical
argument?

Bart

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Krzysztof Kozlowski]

On 21/06/2024 03:14, Lk Sii wrote:
> 
> 
> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
>> Hello:
>>
>> This series was applied to bluetooth/bluetooth-next.git (master)
>> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
>>
> Hi luiz,
> 
> i am curious why Bartosz is able to merge his changes into bluetooth
> development tree bluetooth-next directly.
> 
> 1)
> his changes should belong to *POWER* scope instead of *Bluetooth*
> obviously, however, there are *NOT* any SOB tag from either power and
> bluetooth maintainer. these changes currently only have below Acked-by
> and Signed-off-by tags:

You are trolling us or what?

That's a cross tree pull request.

> 
> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> 2)
> his changes have not merged into linus mainline tree yet.

Do you understand the concept of merge windows?

> 
> 3)
> perhaps, it is safer to pull his changes from linus mainline tree when
> merged than to merge into bluetooth-next firstly.

You are joking, right?


Best regards,
Krzysztof

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Krzysztof Kozlowski]

On 21/06/2024 11:04, Lk Sii wrote:
> On 2024/6/21 14:36, Bartosz Golaszewski wrote:
>> On Fri, Jun 21, 2024 at 3:14 AM Lk Sii <lk_sii@163.com> wrote:
>>>
>>>
>>>
>>> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
>>>> Hello:
>>>>
>>>> This series was applied to bluetooth/bluetooth-next.git (master)
>>>> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
>>>>
>>> Hi luiz,
>>>
>>> i am curious why Bartosz is able to merge his changes into bluetooth
>>> development tree bluetooth-next directly.
>>>
>>
>> This conversation is getting progressively worse...
>>
>>> 1)
>>> his changes should belong to *POWER* scope instead of *Bluetooth*
>>> obviously, however, there are *NOT* any SOB tag from either power and
>>> bluetooth maintainer. these changes currently only have below Acked-by
>>> and Signed-off-by tags:
>>>
>>> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
>>> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>>>
>>
>> It's a new subsystem that has been discussed and reviewed for months
>> and thoroughly tested. Please refer to the cover letter under v8
>> linked in this thread. It's not related to power-management or
>> power-supply, it's its own thing but IMO the best place to put it is
>> under drivers/power/. And I will maintain it.
>>
>>> 2)
>>> his changes have not merged into linus mainline tree yet.
>>>
>>
>> This is why they are in next! They are scheduled to go in during the
>> upcoming merge window. But since changes belong in multiple trees, we
>> need a cross-tree merge.
>>
>>> 3)
>>> perhaps, it is safer to pull his changes from linus mainline tree when
>>> merged than to merge into bluetooth-next firstly.
>>>
>>
>> It's not safer at all, why would spending less time in next be safer?
>>
> it seems this patch serial(new subsystem) does not depend on bluetooth
> and also does not belong to bluetooth subsystem, but have been contained
> by tip of bluetooth tree.
> 
> why not follow below merging produce?
> 1) you send this patch serials to Linus to merge within linus mainline tree
> 2) luiz then pull your changes from linus mainline tree.

This is not how Linux kernel development works. Read process documents
before interfering with people's work.

Best regards,
Krzysztof

Re: [PATCH v9 0/2] pwrseq: introduce the subsystem and first driver

[Luiz Augusto von Dentz]

Hi,

On Fri, Jun 21, 2024 at 5:05 AM Lk Sii <lk_sii@163.com> wrote:
>
> On 2024/6/21 14:36, Bartosz Golaszewski wrote:
> > On Fri, Jun 21, 2024 at 3:14 AM Lk Sii <lk_sii@163.com> wrote:
> >>
> >>
> >>
> >> On 2024/6/20 22:30, patchwork-bot+bluetooth@kernel.org wrote:
> >>> Hello:
> >>>
> >>> This series was applied to bluetooth/bluetooth-next.git (master)
> >>> by Bartosz Golaszewski <bartosz.golaszewski@linaro.org>:
> >>>
> >> Hi luiz,
> >>
> >> i am curious why Bartosz is able to merge his changes into bluetooth
> >> development tree bluetooth-next directly.
> >>
> >
> > This conversation is getting progressively worse...
> >
> >> 1)
> >> his changes should belong to *POWER* scope instead of *Bluetooth*
> >> obviously, however, there are *NOT* any SOB tag from either power and
> >> bluetooth maintainer. these changes currently only have below Acked-by
> >> and Signed-off-by tags:
> >>
> >> Acked-by: Krzysztof Kozlowski <krzk@kernel.org>
> >> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> >>
> >
> > It's a new subsystem that has been discussed and reviewed for months
> > and thoroughly tested. Please refer to the cover letter under v8
> > linked in this thread. It's not related to power-management or
> > power-supply, it's its own thing but IMO the best place to put it is
> > under drivers/power/. And I will maintain it.
> >
> >> 2)
> >> his changes have not merged into linus mainline tree yet.
> >>
> >
> > This is why they are in next! They are scheduled to go in during the
> > upcoming merge window. But since changes belong in multiple trees, we
> > need a cross-tree merge.
> >
> >> 3)
> >> perhaps, it is safer to pull his changes from linus mainline tree when
> >> merged than to merge into bluetooth-next firstly.
> >>
> >
> > It's not safer at all, why would spending less time in next be safer?
> >
> it seems this patch serial(new subsystem) does not depend on bluetooth
> and also does not belong to bluetooth subsystem, but have been contained
> by tip of bluetooth tree.
>
> why not follow below merging produce?
> 1) you send this patch serials to Linus to merge within linus mainline tree
> 2) luiz then pull your changes from linus mainline tree.

This was discussed already, but in case you didn't follow if we need
to incorporate changes earlier we can't really wait the merge window
for the changes to be merged first into the Linus tree. Anyway Bartosz
is the maintainer of Power Sequence subsystem, and several others
drivers, so I think it is safe to trust his judgment here.