This patch introduces one devres API of devres_for_each_res
so that the device's driver can iterate each resource it has
interest in.

The firmware loader will use the API to get each firmware name
from the device instance.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch will store firmware name into devres list of the device
which is requesting firmware loading, so that we can implement
auto cache and uncache firmware for devices in need.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

request_firmware_nowait is allowed to be called in atomic
context now if @gfp is GFP_ATOMIC, so fix the obsolete
comments and states which situations are suitable for using
it.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Callers of request_firmware* must hold the reference count of
@device, otherwise it is easy to trigger oops since the firmware
loader device is the child of @device.

This patch adds comments about the usage. In fact, most of drivers
call request_firmware* in its probe() or open(), so the constraint
should be reasonable and can be satisfied.

Also this patch holds the reference count of @device before
schedule_work() in request_firmware_nowait() to avoid that
the @device is released after request_firmware_nowait returns
and before the worker function is scheduled.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patches introduce two kernel APIs of cache_firmware and
uncache_firmware, both of which take the firmware file name
as the only parameter.

So any drivers can call cache_firmware to cache the specified
firmware file into kernel memory, and can use the cached firmware
in situations which can't request firmware from user space.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch always let firmware_buf own the pages buffer allocated
inside firmware_data_write, and add all instances of firmware_buf
into the firmware cache global list. Also introduce one private field
in 'struct firmware', so release_firmware will see the instance of
firmware_buf associated with the current firmware instance, then just
'free' the instance of firmware_buf.

The firmware_buf instance represents one pages buffer for one
firmware image, so lots of firmware loading requests can share
the same firmware_buf instance if they request the same firmware
image file.

This patch will make implementation of the following cache_firmware/
uncache_firmware very easy and simple.

In fact, the patch improves request_formware/release_firmware:

        - only request userspace to write firmware image once if
	several devices share one same firmware image and its drivers
	call request_firmware concurrently.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch introduces struct firmware_buf to describe the buffer
which holds the firmware data, which will make the following
cache_firmware/uncache_firmware implemented easily.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

If one device driver calls request_firmware_nowait() to request
several different firmwares' loading, device_add() will return
failure since all firmware loader device use same name of the
device who is requesting firmware.

This patch always use the name of firmware image as the firmware
loader device name to fix the problem since the following patches
for caching firmware will make sure only one loading for same
firmware is alllowd at the same time.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

The wmb() inside fw_load_abort is not necessary, since
complete() and wait_on_completion() has implied one pair
of memory barrier.

Also wmb() isn't a correct usage, so just remove it.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch fixes two races in loading firmware:

1, FW_STATUS_DONE should be set before waking up the task waitting
on _request_firmware_load, otherwise FW_STATUS_ABORT may be
thought as DONE mistakenly.

2, Inside _request_firmware_load(), there is a small window between
wait_for_completion() and mutex_lock(&fw_lock), and 'echo 1 > loading'
still may happen during the period, so this patch checks FW_STATUS_DONE
to prevent pages' buffer completed from being freed in firmware_loading_store.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch doesn't transfer ownership of pages' buffer to the
instance of firmware until the firmware loading is completed,
which will simplify firmware_loading_store a lot, so help
to introduce the following cache_firmware and uncache_firmware
mechanism during system suspend-resume cycle.

In fact, this patch fixes one bug: if writing data into
firmware loader device is bypassed between writting 1 and 0 to
'loading', OOPS will be triggered without the patch.

Also handle the vmap failure case, and add some comments to make
code more readable.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Right now we have support for explicit platform device IDs, as well as
ID-less platform devices when a given device type can only have one
instance. However there are cases where multiple instances of a device
type can exist, and their IDs aren't (and can't be) known in advance
and do not matter. In that case we need automatic device IDs to avoid
device name collisions.

I am using magic ID value -2 (PLATFORM_DEVID_AUTO) for this, similar
to -1 for ID-less devices. The automatically allocated device IDs are
global (to avoid an additional per-driver cost.) We keep note that the
ID was automatically allocated so that it can be freed later.

Note that we also restore the ID to PLATFORM_DEVID_AUTO on error and
device deletion, to avoid avoid unexpected behavior on retry. I don't
really expect retries on platform device addition, but better safe
than sorry.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

device_del can happen anytime, so once it happens,
the devres of the device will be freed inside device_del, but
drivers can't know it has been deleted and may still add
resources into the device, so memory leak is caused.

This patch moves the devres_release_all to fix the problem.

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/page_alloc.c has some memory isolation functions but they are used only
when we enable CONFIG_{CMA|MEMORY_HOTPLUG|MEMORY_FAILURE}.  So let's make
it configurable by new CONFIG_MEMORY_ISOLATION so that it can reduce
binary size and we can check it simple by CONFIG_MEMORY_ISOLATION, not if
defined CONFIG_{CMA|MEMORY_HOTPLUG|MEMORY_FAILURE}.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

This patch adds dma_get_sgtable() function which is required to let
drivers to share the buffers allocated by DMA-mapping subsystem. Right
now the driver gets a dma address of the allocated buffer and the kernel
virtual mapping for it. If it wants to share it with other device (= map
into its dma address space) it usually hacks around kernel virtual
addresses to get pointers to pages or assumes that both devices share
the DMA address space. Both solutions are just hacks for the special
cases, which should be avoided in the final version of buffer sharing.

To solve this issue in a generic way, a new call to DMA mapping has been
introduced - dma_get_sgtable(). It allocates a scatter-list which
describes the allocated buffer and lets the driver(s) to use it with
other device(s) by calling dma_map_sg() on it.

This patch provides a generic implementation based on virt_to_page()
call. Architectures which require more sophisticated translation might
provide their own get_sgtable() methods.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reviewed-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Commit 9adc5374 ('common: dma-mapping: introduce mmap method') added a
generic method for implementing mmap user call to dma_map_ops structure.

This patch converts ARM and PowerPC architectures (the only providers of
dma_mmap_coherent/dma_mmap_writecombine calls) to use this generic
dma_map_ops based call and adds a generic cross architecture
definition for dma_mmap_attrs, dma_mmap_coherent, dma_mmap_writecombine
functions.

The generic mmap virt_to_page-based fallback implementation is provided for
architectures which don't provide their own implementation for mmap method.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reviewed-by: Kyungmin Park <kyungmin.park@samsung.com>

releases what needs to be released after {kern,user}_path_create()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Now that scsi registers its async scan work with the async subsystem,
wait_for_device_probe() is sufficient for ensuring all scanning is
complete.

[jejb: fix merge problems with eea03c20 Make wait_for_device_probe() also do scsi_complete_async_scans()]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Eldad Zack <eldad@fogrefinery.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

Commit cf579dfb (PM / Sleep: Introduce
"late suspend" and "early resume" of devices) introduced a bug where
suspend_late handlers would be called, but if dpm_suspend_noirq returned
an error the early_resume handlers would never be called.  All devices
would end up on the dpm_late_early_list, and would never be resumed
again.

Fix it by calling dpm_resume_early when dpm_suspend_noirq returns
an error.

Signed-off-by: Colin Cross <ccross@android.com>
Cc: stable@vger.kernel.org
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Commit a7a20d10 ("sd: limit the scope of the async probe domain")
make the SCSI device probing run device discovery in it's own async
domain.

However, as a result, the partition detection was no longer synchronized
by async_synchronize_full() (which, despite the name, only synchronizes
the global async space, not all of them).  Which in turn meant that
"wait_for_device_probe()" would not wait for the SCSI partitions to be
parsed.

And "wait_for_device_probe()" was what the boot time init code relied on
for mounting the root filesystem.

Now, most people never noticed this, because not only is it
timing-dependent, but modern distributions all use initrd.  So the root
filesystem isn't actually on a disk at all.  And then before they
actually mount the final disk filesystem, they will have loaded the
scsi-wait-scan module, which not only does the expected
wait_for_device_probe(), but also does scsi_complete_async_scans().

[ Side note: scsi_complete_async_scans() had also been partially broken,
  but that was fixed in commit 43a8d39d ("fix async probe
  regression"), so that same commit a7a20d10 had actually broken
  setups even if you used scsi-wait-scan explicitly ]

Solve this problem by just moving the scsi_complete_async_scans() call
into wait_for_device_probe().  Everybody who wants to wait for device
probing to finish really wants the SCSI probing to complete, so there's
no reason not to do this.

So now "wait_for_device_probe()" really does what the name implies, and
properly waits for device probing to finish.  This also removes the now
unnecessary extra calls to scsi_complete_async_scans().

Reported-and-tested-by: Artem S. Tashkinov <t.artem@mailcity.com>
Cc: Dan Williams <dan.j.williams@gmail.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: James Bottomley <jbottomley@parallels.com>
Cc: Borislav Petkov <bp@amd64.org>
Cc: linux-scsi <linux-scsi@vger.kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Fix the following sparse warning:
drivers/base/power/qos.c:465:29: warning: Using plain integer as NULL pointer

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Fix the following sparse warnings:
drivers/base/power/main.c:48:1: warning: symbol 'dpm_prepared_list' was not declared. Should it be static?
drivers/base/power/main.c:49:1: warning: symbol 'dpm_suspended_list' was not declared. Should it be static?
drivers/base/power/main.c:50:1: warning: symbol 'dpm_late_early_list' was not declared. Should it be static?
drivers/base/power/main.c:51:1: warning: symbol 'dpm_noirq_list' was not declared. Should it be static?

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Signed-off-by: Dimitris Papastamos <dp@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Do not send the uevent if driver_add_groups failed.

Reported-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

The pm_runtime_get_noresume() calls before really_probe() and
before executing __device_attach() for each driver on the
device's bus cause problems to happen if probing fails and if the
driver has enabled runtime PM for the device in its .probe()
callback.  Namely, in that case, if the device has been resumed
by the driver after enabling its runtime PM and if it turns out that
.probe() should return an error, the driver is supposed to suspend
the device and disable its runtime PM before exiting .probe().
However, because the device's runtime PM usage counter was
incremented by the core before calling .probe(), the driver's attempt
to suspend the device will not succeed and the device will remain in
the full-power state after the failing .probe() has returned.

To fix this issue, remove the pm_runtime_get_noresume() calls from
driver_probe_device() and from device_attach() and replace the
corresponding pm_runtime_put_sync() calls with pm_runtime_idle()
to preserve the existing behavior (which is to check if the device
is idle and to suspend it eventually in that case after probing).

Reported-and-tested-by: Kevin Hilman <khilman@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

When deferred probe was originally added the idea was that devices which
defer their probes would move themselves to the end of dpm_list in order
to try to keep the assumptions that we're making about the list being in
roughly the order things should be suspended correct. However this hasn't
been what's been happening and doing it requires a lot of duplicated code
to do the moves.

Instead take a simple, brute force solution and have the deferred probe
code push devices to the end of dpm_list before it retries the probe. This
does mean we lock the dpm_list a bit more often but it's very simple and
the code shouldn't be a fast path. We do the move with the deferred mutex
dropped since doing things with fewer locks held simultaneously seems like
a good idea.

This approach was most recently suggested by Grant Likely.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Grant Likely <grant.likely@secretlab.ca>,
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Device driver attribute groups are created after userspace is notified
via an add event. Fix this by moving the kobject_uevent call to
driver_register after the attribute groups are added.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Firstly, .shutdown callback may touch a uninitialized hardware
if dev->driver is set and .probe is not completed.

Secondly, device_shutdown() may dereference a null pointer to cause
oops when dev->driver is cleared after it has been checked in
device_shutdown().

So just hold device lock and its parent lock(if it has) to
fix the races.

Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

all callers want the same thing, actually - a kinda-sorta analog of
kern_path_create().  I.e. they want parent vfsmount/dentry (with
->i_mutex held, to make sure the child dentry is still their child)
+ the child dentry.

Signed-off-by Al Viro <viro@zeniv.linux.org.uk>

The power/async device sysfs attribute is only used if both
CONFIG_PM_ADVANCED_DEBUG and CONFIG_PM_SLEEP are set, but the code
implementing it doesn't depend on CONFIG_PM_SLEEP.  As a result, a
build warning appears if CONFIG_PM_ADVANCED_DEBUG is set and
CONFIG_PM_SLEEP is not set.

Fix it by adding a #ifdef CONFIG_PM_SLEEP around the code in
question.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

The functions genpd_save_dev() and genpd_restore_dev() are not used
for CONFIG_PM_RUNTIME unset, so move them under an appropriate
#ifdef.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Fixes the following sparse warning:
drivers/base/power/domain.c:1679:55: warning: Using plain integer as NULL pointer

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Fixes the folloiwng sparse warning:
drivers/base/power/domain.c:149:5:
warning: symbol '__pm_genpd_poweron' was not declared. Should it be static?

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

On certain bios, resume hangs if cpus are allowed to enter idle states
during suspend [1].

This was fixed in apci idle driver [2].But intel_idle driver does not
have this fix. Thus instead of replicating the fix in both the idle
drivers, or in more platform specific idle drivers if needed, the
more general cpuidle infrastructure could handle this.

A suspend callback in cpuidle_driver could handle this fix. But
a cpuidle_driver provides only basic functionalities like platform idle
state detection capability and mechanisms to support entry and exit
into CPU idle states. All other cpuidle functions are found in the
cpuidle generic infrastructure for good reason that all cpuidle
drivers, irrepective of their platforms will support these functions.

One option therefore would be to register a suspend callback in cpuidle
which handles this fix. This could be called through a PM_SUSPEND_PREPARE
notifier. But this is too generic a notfier for a driver to handle.

Also, ideally the job of cpuidle is not to handle side effects of suspend.
It should expose the interfaces which "handle cpuidle 'during' suspend"
or any other operation, which the subsystems call during that respective
operation.

The fix demands that during suspend, no cpus should be allowed to enter
deep C-states. The interface cpuidle_uninstall_idle_handler() in cpuidle
ensures that. Not just that it also kicks all the cpus which are already
in idle out of their idle states which was being done during cpu hotplug
through a CPU_DYING_FROZEN callbacks.

Now the question arises about when during suspend should
cpuidle_uninstall_idle_handler() be called. Since we are dealing with
drivers it seems best to call this function during dpm_suspend().
Delaying the call till dpm_suspend_noirq() does no harm, as long as it is
before cpu_hotplug_begin() to avoid race conditions with cpu hotpulg
operations. In dpm_suspend_noirq(), it would be wise to place this call
before suspend_device_irqs() to avoid ugly interactions with the same.

Ananlogously, during resume.

References:
[1] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/674075.
[2] http://marc.info/?l=linux-pm&m=133958534231884&w=2



Reported-and-tested-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Sometimes for failures during very early init the trace infrastructure
isn't available early enough to be used.  For this sort of problem
defining LOG_DEVICE will add printks for basic register I/O on a specific
device, allowing trace to be extracted when the trace system doesn't come
up early enough to work with.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Make it possible to modify device callbacks used by the generic PM
domains core code at any time, not only after the device has been
added to a domain.  This will allow device drivers to provide their
own device PM domain callbacks even if they are registered before
adding the devices to PM domains.

For this purpose, use the observation that the struct
generic_pm_domain_data object containing the relevant callback
pointers may be allocated by pm_genpd_add_callbacks() and the
callbacks may be set before __pm_genpd_add_device() is run for
the given device.  This object will then be used by
__pm_genpd_add_device(), but it has to be protected from
premature removal by reference counting.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Add a mechanism for counting references to the
struct generic_pm_domain_data object pointed to by
dev->power.subsys_data->domain_data if the device in question
belongs to a generic PM domain.

This change is necessary for a subsequent patch making it possible to
allocate that object from within pm_genpd_add_callbacks(), so that
drivers can attach their PM domain device callbacks to devices before
those devices are added to PM domains.

This patch has been tested on the SH7372 Mackerel board.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

On some systems there are CPU cores located in the same power
domains as I/O devices.  Then, power can only be removed from the
domain if all I/O devices in it are not in use and the CPU core
is idle.  Add preliminary support for that to the generic PM domains
framework.

First, the platform is expected to provide a cpuidle driver with one
extra state designated for use with the generic PM domains code.
This state should be initially disabled and its exit_latency value
should be set to whatever time is needed to bring up the CPU core
itself after restoring power to it, not including the domain's
power on latency.  Its .enter() callback should point to a procedure
that will remove power from the domain containing the CPU core at
the end of the CPU power transition.

The remaining characteristics of the extra cpuidle state, referred to
as the "domain" cpuidle state below, (e.g. power usage, target
residency) should be populated in accordance with the properties of
the hardware.

Next, the platform should execute genpd_attach_cpuidle() on the PM
domain containing the CPU core.  That will cause the generic PM
domains framework to treat that domain in a special way such that:

 * When all devices in the domain have been suspended and it is about
   to be turned off, the states of the devices will be saved, but
   power will not be removed from the domain.  Instead, the "domain"
   cpuidle state will be enabled so that power can be removed from
   the domain when the CPU core is idle and the state has been chosen
   as the target by the cpuidle governor.

 * When the first I/O device in the domain is resumed and
   __pm_genpd_poweron(() is called for the first time after
   power has been removed from the domain, the "domain" cpuidle
   state will be disabled to avoid subsequent surprise power removals
   via cpuidle.

The effective exit_latency value of the "domain" cpuidle state
depends on the time needed to bring up the CPU core itself after
restoring power to it as well as on the power on latency of the
domain containing the CPU core.  Thus the "domain" cpuidle state's
exit_latency has to be recomputed every time the domain's power on
latency is updated, which may happen every time power is restored
to the domain, if the measured power on latency is greater than
the latency stored in the corresponding generic_pm_domain structure.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reviewed-by: Kevin Hilman <khilman@ti.com>

While resuming a device belonging to a PM domain,
pm_genpd_runtime_resume() calls __pm_genpd_restore_device() to
restore its state, if necessary.  The latter starts the device,
using genpd_start_dev(), restores its state, using
genpd_restore_dev(), and then stops it, using genpd_stop_dev().
However, this last operation is not necessary, because the
device is supposed to be operational after pm_genpd_runtime_resume()
has returned and because of it pm_genpd_runtime_resume() has to
call genpd_start_dev() once again for the "restored" device, which
is inefficient.

To make things more efficient, remove the call to genpd_stop_dev()
from __pm_genpd_restore_device() and the direct call to
genpd_start_dev() from pm_genpd_runtime_resume().  [Of course,
genpd_start_dev() still has to be called by it for devices with the
power.irq_safe flag set, because __pm_genpd_restore_device() is not
executed for them.]

This change has been tested on the SH7372 Mackerel board.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>