Measure latency does by itself contribute to an increased latency, thus we
should avoid it when it isn't needed.

By merging the latency measurements for the ->save_state() and the
->stop() callbacks, we get one measurement instead of two and we get one
value to store instead of two. Let's also apply the likewise change for
the ->start() and ->restore_state() callbacks.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Lina Iyer <lina.iyer@linaro.org>
Acked-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Once genpd could be configured to be built with CONFIG_PM_RUNTIME unset
(nowadays CONFIG_PM), the pm_genpd_poweron() API served a purpose, since
it allowed users to power on a PM domain.

As such configuration no longer is supported, users shall solely rely on
using some of the runtime PM APIs to power on a PM domain.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

As the last user of the pm_genpd_poweroff_unused() API has moved into
relying on genpd to deal with this internally from a late_initcall, let's
remove the API.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Commit ba2bbfbf6307 ("PM / Domains: Remove intermediate states..") changed
the power off sequence (pm_genpd_poweroff()), which from locking point of
view means the genpd mutex is held throughout the sequence.

The above change means the in_progress counter can't be updated while
pm_genpd_poweroff() is executing, which allows us to remove the counter.

Instead we inform pm_genpd_poweroff() via a bool parameter, to indicate
whether we call it from the scheduled work or from the ->runtime_suspend()
callback, since that all that matters.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Lina Iyer <lina.iyer@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

The power domains code allows to tie a cpuidle state with a power domain.

Preventing the cpuidle framework to enter a specific idle state by disabling
from the power domain framework is a good idea. Unfortunately, the current
implementation has some gaps with a SMP system and a complex cpuidle
implementation. Enabling a power domain wakes up all the cpus even if a cpu
does not belong to the power domain.

There is some work to do a logical representation with the power domains of
the hardware dependencies (eg. a cpu belongs to a power domains, these power
domains belong to a higher power domain for a cluster, etc ...). A new code
relying on the genpd hierarchy to disable the idle states would make more
sense.

As the unique user of this code has been removed, let's wipe out this code
to prevent new user and to have a clean place to put a new implementation.

Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Acked-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

As all users of the named based APIs now have converted to the non-named
based APIs, the time has come to remove them.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

The governor dummies for the !CONFIG_PM_GENERIC_DOMAINS case are
unusable, as a governors is always referred to by taking its address,
which you can't do with a literal NULL pointer.

I.e.

	pm_genpd_init(genpd, &simple_qos_governor, false);

fails to compile with:

	error: lvalue required as unary '&' operand

Hence just remove the governor dummies.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Genpd's ->runtime_suspend() (assigned to pm_genpd_runtime_suspend())
doesn't immediately walk the hierarchy of ->runtime_suspend() callbacks.
Instead, pm_genpd_runtime_suspend() calls pm_genpd_poweroff() which
postpones that until *all* the devices in the genpd are runtime suspended.

When pm_genpd_poweroff() discovers that the last device in the genpd is
about to be runtime suspended, it calls __pm_genpd_save_device() for *all*
the devices in the genpd sequentially. Furthermore,
__pm_genpd_save_device() invokes the ->start() callback, walks the
hierarchy of the ->runtime_suspend() callbacks and invokes the ->stop()
callback. This causes a "thundering herd" problem.

Let's address this issue by having pm_genpd_runtime_suspend() immediately
walk the hierarchy of the ->runtime_suspend() callbacks, instead of
postponing that to the power off sequence via pm_genpd_poweroff(). If the
selected ->runtime_suspend() callback doesn't return an error code, call
pm_genpd_poweroff() to see if it's feasible to also power off the PM
domain.

Adopting this change enables us to simplify parts of the code in genpd,
for example the locking mechanism. Additionally, it gives some positive
side effects, as described below.

i)
One device's ->runtime_resume() latency is no longer affected by other
devices' latencies in a genpd.

The complexity genpd has to support the option to abort the power off
sequence suffers from latency issues. More precisely, a device that is
requested to be runtime resumed, may end up waiting for
__pm_genpd_save_device() to complete its operations for *another* device.
That's because pm_genpd_poweroff() can't confirm an abort request while it
waits for __pm_genpd_save_device() to return.

As this patch removes the intermediate states in pm_genpd_poweroff() while
powering off the PM domain, we no longer need the ability to abort that
sequence.

ii)
Make pm_runtime[_status]_suspended() reliable when used with genpd.

Until the last device in a genpd becomes idle, pm_genpd_runtime_suspend()
will return 0 without actually walking the hierarchy of the
->runtime_suspend() callbacks. However, by returning 0 the runtime PM core
considers the device as runtime_suspended, so
pm_runtime[_status]_suspended() will return true, even though the device
isn't (yet) runtime suspended.

After this patch, since pm_genpd_runtime_suspend() immediately walks the
hierarchy of the ->runtime_suspend() callbacks,
pm_runtime[_status]_suspended() will accurately reflect the status of the
device.

iii)
Enable fine-grained PM through runtime PM callbacks in drivers/subsystems.

There are currently cases were drivers/subsystems implements runtime PM
callbacks to deploy fine-grained PM (e.g. gate clocks, move pinctrl to
power-save state, etc.). While using the genpd, pm_genpd_runtime_suspend()
postpones invoking these callbacks until *all* the devices in the genpd
are runtime suspended. In essence, one runtime resumed device prevents
fine-grained PM for other devices within the same genpd.

After this patch, since pm_genpd_runtime_suspend() immediately walks the
hierarchy of the ->runtime_suspend() callbacks, fine-grained PM is enabled
throughout all the levels of runtime PM callbacks.

iiii)
Enable fine-grained PM for IRQ safe devices

Per the definition for an IRQ safe device, its runtime PM callbacks must
be able to execute in atomic context. In the path while genpd walks the
hierarchy of the ->runtime_suspend() callbacks for the device, it uses a
mutex. Therefore, genpd prevents that path to be executed for IRQ safe
devices.

As this patch changes pm_genpd_runtime_suspend() to immediately walk the
hierarchy of the ->runtime_suspend() callbacks and without needing to use
a mutex, fine-grained PM is enabled throughout all the levels of runtime
PM callbacks for IRQ safe devices.

Unfortunately this patch also comes with a drawback, as described in the
summary below.

Driver's/subsystem's runtime PM callbacks may be invoked even when the
genpd hasn't actually powered off the PM domain, potentially introducing
unnecessary latency.

However, in most cases, saving/restoring register contexts for devices are
typically fast operations or can be optimized in device specific ways
(e.g. shadow copies of register contents in memory, device-specific checks
to see if context has been lost before restoring context, etc.).

Still, in some cases the driver/subsystem may suffer from latency if
runtime PM is used in a very fine-grained manner (e.g. for each IO request
or xfer). To prevent that extra overhead, the driver/subsystem may deploy
the runtime PM autosuspend feature.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Tested-by: Lina Iyer <lina.iyer@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

The PM domain code contains two methods to get the generic PM domain
for a struct device.  One is dev_to_genpd() which is only safe when
we know for certain that the device has a generic PM domain attached.
The other is coded into genpd_dev_pm_detach() which ensures that the
PM domain in the struct device is a generic PM domain (and so is safer).

This commit factors out the safer version, documents it, and hides the
unsafe dev_to_genpd().

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

While adding devices to their PM domains, dev_pm_qos_add_notifier() was
invoked while allocating the generic_pm_domain_data for the device.

Since the generic_pm_domain_data's device pointer will be assigned
after allocation, the ->genpd_dev_pm_qos_notifier() callback could be
called prior having a valid pointer to the device. Similar scenario
existed while removing a device from a genpd.

To cope with these scenarios a mutex was used to protect the pointer to
the device.

By re-order the sequence for when dev_pm_qos_add|remove_notifier() are
invoked, we make sure the ->genpd_dev_pm_qos_notifier() callback are
always called with a valid device pointer available.

In this way, we eliminate the need for protecting the pointer and thus
we can remove the mutex from the struct generic_pm_domain_data.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>