There are drivers that can not be probed asynchronously. One such group
is platform drivers registered with platform_driver_probe(), which
expects driver's probe routine be discarded after the driver has been
registered and initial binding attempt executed. Also
platform_driver_probe() an error when no devices were bound to the
driver, allowing failing to load such driver module altogether.

Other drivers do not work well with asynchronous probing because of
driver bug or not optimal driver organization.

To allow using such drivers even when user requests asynchronous probing
as default boot strategy, let's allow them to opt out.
Signed-off-by: NLuis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Some init systems may wish to express the desire to have device drivers
run their probe() code asynchronously. This implements support for this
and allows userspace to request async probe as a preference through a
generic shared device driver module parameter, async_probe.

Implementation for async probe is supported through a module parameter
given that since synchronous probe has been prevalent for years some
userspace might exist which relies on the fact that the device driver
will probe synchronously and the assumption that devices it provides
will be immediately available after this.
Signed-off-by: NLuis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Some devices take a long time when initializing, and not all drivers are
suited to initialize their devices when they are open. For example,
input drivers need to interrogate their devices in order to publish
device's capabilities before userspace will open them. When such drivers
are compiled into kernel they may stall entire kernel initialization.

This change allows drivers request for their probe functions to be
called asynchronously during driver and device registration (manual
binding is still synchronous). Because async_schedule is used to perform
asynchronous calls module loading will still wait for the probing to
complete.

Note that the end goal is to make the probing asynchronous by default,
so annotating drivers with PROBE_PREFER_ASYNCHRONOUS is a temporary
measure that allows us to speed up boot process while we validating and
fixing the rest of the drivers and preparing userspace.

This change is based on earlier patch by "Luis R. Rodriguez"
<mcgrof@suse.com>
Signed-off-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Currently there is no way to query which CPUs are in nohz_full
mode from userspace.

Export the CPU list running in nohz_full mode in sysfs,
specifically in the file /sys/devices/system/cpu/nohz_full

This can be used by system management tools like libvirt,
openstack, and others to ensure proper task placement.
Signed-off-by: NRik van Riel <riel@redhat.com>
Acked-by: NMike Galbraith <umgwanakikbuti@gmail.com>
Acked-by: NChris Metcalf <cmetcalf@ezchip.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

After system bootup, there is no totally reliable way to see
which CPUs are isolated, because the kernel may modify the
CPUs specified on the isolcpus= kernel command line option.

Export the CPU list that actually got isolated in sysfs,
specifically in the file /sys/devices/system/cpu/isolated

This can be used by system management tools like libvirt,
openstack, and others to ensure proper placement of tasks.
Suggested-by: NLi Zefan <lizefan@huawei.com>
Signed-off-by: NRik van Riel <riel@redhat.com>
Acked-by: NMike Galbraith <umgwanakikbuti@gmail.com>
Acked-by: NChris Metcalf <cmetcalf@ezchip.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The seq_printf return value, because it's frequently misused,
will eventually be converted to void.

See: commit 1f33c41c ("seq_file: Rename seq_overflow() to
     seq_has_overflowed() and make public")
Signed-off-by: NJoe Perches <joe@perches.com>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Len Brown <len.brown@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NDavid Howells <dhowells@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

There's a deadlock when concurrently hot-adding memory through the probe
interface and switching a memory block from offline to online.

When hot-adding memory via the probe interface, add_memory() first takes
mem_hotplug_begin() and then device_lock() is later taken when registering
the newly initialized memory block.  This creates a lock dependency of (1)
mem_hotplug.lock (2) dev->mutex.

When switching a memory block from offline to online, dev->mutex is first
grabbed in device_online() when the write(2) transitions an existing
memory block from offline to online, and then online_pages() will take
mem_hotplug_begin().

This creates a lock inversion between mem_hotplug.lock and dev->mutex.
Vitaly reports that this deadlock can happen when kworker handling a probe
event races with systemd-udevd switching a memory block's state.

This patch requires the state transition to take mem_hotplug_begin()
before dev->mutex.  Hot-adding memory via the probe interface creates a
memory block while holding mem_hotplug_begin(), there is no way to take
dev->mutex first in this case.

online_pages() and offline_pages() are only called when transitioning
memory block state.  We now require that mem_hotplug_begin() is taken
before calling them -- this requires exporting the mem_hotplug_begin() and
mem_hotplug_done() to generic code.  In all hot-add and hot-remove cases,
mem_hotplug_begin() is done prior to device_online().  This is all that is
needed to avoid the deadlock.
Signed-off-by: NDavid Rientjes <rientjes@google.com>
Reported-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Tested-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Zhang Zhen <zhenzhang.zhang@huawei.com>
Cc: Vladimir Davydov <vdavydov@parallels.com>
Cc: Wang Nan <wangnan0@huawei.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use macro section_nr_to_pfn() to switch between section and pfn, instead
of open-coding it.  No semantic changes.
Signed-off-by: NSheng Yong <shengyong1@huawei.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Introduce data structures and code allowing "built-in" properties
to be associated with devices in such a way that they will be used
by the device_property_* API if no proper firmware node (neither DT
nor ACPI) is present for the given device.

Each property is to be represented by a property_entry structure.
An array of property_entry structures (terminated with a null
entry) can be pointed to by the properties field of struct
property_set that can be added as a firmware node to a struct
device using device_add_property_set().  That will cause the
device_property_* API to use that property_set as the source
of properties if the given device does not have a DT node or
an ACPI companion device object associated with it.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: NHeikki Krogerus <heikki.krogerus@linux.intel.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Add a secondary pointer to struct fwnode_handle so as to make it
possible for a device to have two firmware nodes associated with
it at the same time, for example, an ACPI node and a node with
a set of properties provided by platform initialization code.

In the future that will allow device property lookup to fall back
from the primary firmware node to the secondary one if the given
property is not present there to make it easier to provide defaults
for device properties used by device drivers.

Introduce two helper routines, set_primary_fwnode() and
set_secondary_fwnode() allowing callers to add a primary/secondary
firmware node to the given device in such a way that

 (1) If there's only one firmware node for that device, it will be
     pointed to by the device's firmware node pointer.
 (2) If both the primary and secondary firmware nodes are present,
     the primary one will be pointed to by the device's firmware
     node pointer, while the secondary one will be pointed to by the
     primary node's secondary pointer.
 (3) If one of these nodes is removed (by calling one of the new
     nelpers with NULL as the second argument), the other one will
     be preserved.

Make ACPI use set_primary_fwnode() for attaching its firmware nodes
to devices.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: NHeikki Krogerus <heikki.krogerus@linux.intel.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Now that the ACPI companions of devices are pointed to by the fwnode
field in struct device, the device_property_*() accessor functions
can be modified to use their fwnode_property_*() counterparts
internally with minimum extra overhead in the IS_ENABLED(CONFIG_OF)
case, so make those changes.

This allows us to get rid of the rather ugly DEV_PROP_READ_ARRAY()
macro among other things.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Commit 5c0acf3b (driver core: Add comments about returning array
counts) forgot to update fwnode_property_read_string_array() along
the lines of device_property_read_string_array(), although it did
change the kerneldoc comment of it.  Fix that.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This fixes a regression from the net subsystem:
After commit d52fdbb7
"smc91x: retrieve IRQ and trigger flags in a modern way"
a regression would appear on some legacy platforms such
as the ARM PXA Zylonite that specify IRQ resources like
this:

static struct resource r = {
       .start  = X,
       .end    = X,
       .flags  = IORESOURCE_IRQ | IORESOURCE_IRQ_HIGHEDGE,
};

The previous code would retrieve the resource and parse
the high edge setting in the SMC91x driver, a use pattern
that means every driver specifying an IRQ flag from a
static resource need to parse resource flags and apply
them at runtime.

As we switched the code to use IRQ descriptors to retrieve
the the trigger type like this:

  irqd_get_trigger_type(irq_get_irq_data(...));

the code would work for new platforms using e.g. device
tree as the backing irq descriptor would have its flags
properly set, whereas this kind of oldstyle static
resources at no point assign the trigger flags to the
corresponding IRQ descriptor.

To make the behaviour identical on modern device tree
and legacy static platform data platforms, modify
platform_get_irq() to assign the trigger flags to the
irq descriptor when a client looks up an IRQ from static
resources.

Fixes: d52fdbb7 ("smc91x: retrieve IRQ and trigger flags in a modern way")
Tested-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Currently probe deferral prints a message every time a device requests
deferral at info severity (which is displayed by default). This can have
an impact on system boot times with serial consoles and is generally quite
noisy.

Since subsystems and drivers should already be logging the specific reason
for probe deferral in order to aid users in understanding problems the
messages from the driver core should be redundant lower the severity of
the messages printed, cutting down on the volume of output on the console.

This does mean that if the drivers and subsystems aren't doing a good job
we get no output on the console by default. Ideally we'd be able to arrange
to print if nothing else printed, though that's a little fun. Even better
would be to come up with a mechanism that explicitly does dependencies so
we don't have to keep polling and erroring.
Signed-off-by: NMark Brown <broonie@kernel.org>
Reviewed-by: NBjorn Andersson <bjorn.andersson@sonymobile.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

So I've been annoyed lately with having a bunch of devices such as i2c
eeproms (for use by VPDs, server world !) and other bits and pieces that
I want to be able to identify from userspace, and possibly provide
additional data about from FW.

Basically, it boils down to correlating the sysfs device with the OF
tree device node, so that user space can use device-tree info such as
additional "location" or "label" (or whatever else we can come up with)
propreties to identify a given device, or get some attributes of use
about it, etc...

Now, so far, we've done that in some subsystem in a fairly ad-hoc basis
using "devspec" properties. For example, PCI creates them if it can
correlate the probed device with a DT node. Some powerpc specific busses
do that too.

However, i2c doesn't and it would be nice to have something more generic
since technically any device can have a corresponding device tree node.

This patch adds an "of_node" symlink to devices that have a non-NULL
dev->of_node pointer, the patch is pretty trivial and seems to work just
fine for me.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: NRob Herring <robh@kernel.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

When using the user mode helper to load firmwares the function _request_firmware
gets a positive return value from fw_load_from_user_helper and because of this
the firmware buffer is not assigned. This happens only when the return value
is zero. This patch fixes this problem in _request_firmware_load. When the
completion is ready the return value is set to zero.
Signed-off-by: NZahari Doychev <zahari.doychev@linux.com>
Cc: Ming Lei <ming.lei@canonical.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use the static attribute groups assigned to the device instead of
manual device_create_file() & co calls.  It simplifies the code and
can avoid possible races, too.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

On architectures that depend on DT for obtaining cache hierarcy, we need
to validate the device node for all the cache indices, failing to do so
might result in wrong information being exposed to the userspace.

This is quite possible on initial/incomplete versions of the device
trees. In such cases, it's better to bail out if all the required device
nodes are not present.

This patch adds checks for the validation of device node for all the
caches and doesn't initialise the cacheinfo if there's any error.
Reported-by: NMark Rutland <mark.rutland@arm.com>
Acked-by: NMark Rutland <mark.rutland@arm.com>
Signed-off-by: NSudeep Holla <sudeep.holla@arm.com>
Cc: stable <stable@vger.kernel.org> # 4.0
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Linux kernel coding style require that tabs should be used instead of
spaces for code indentation.
Problem found using checkpatch.pl script.
Signed-off-by: NLavinia Tache <lavinia.tachee@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Found using checkpatch.pl
Signed-off-by: NLavinia Tache <lavinia.tachee@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch fixes the following error found by checkpatch.pl:
ERROR: "foo * bar" should be "foo *bar"
Signed-off-by: NAna Nedelcu <anafnedelcu@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch changes spaces to tabs. Found using checkpatch.pl
Signed-off-by: NIoana Ciornei <ciorneiioana@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Fix checkpatch.pl issues with coding style. Removed whitespace and
fixed indentation
Signed-off-by: NAndrei Oprea <andrei.br92@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch fixes the following checkpatch.pl error:
ERROR: "foo * bar" should be "foo *bar"
Signed-off-by: NMarius Cristian Eseanu <eseanu.cristian@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch fixes the following warning found by checkpatch.pl:
WARNING: Missing a black line after declarations
Signed-off-by: NCosmin Tomulescu <cosmintom@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Found with checkpatch.pl
Signed-off-by: NCosmin Dragomir <cosmin.gabriel.dragomir@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Signed-off-by: NIoana Ciornei <ciorneiioana@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Reported by checkpatch.pl

While at it, removed blank line between function call and error
checking.
Signed-off-by: NAndrei Poenaru <andreigpoenaru@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Found this using checkpatch.pl.
Signed-off-by: NFlorin Papa <florin.papa192@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The put_device() function tests whether its argument is NULL and then
returns immediately. Thus the test around the call is not needed.

This issue was detected by using the Coccinelle software.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

We can use the ATTRIBUTE_GROUPS() macro here, so use it, saving some
lines of code.

Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org

Instead of manual calls of multiple device_create_file() and
device_remove_file(), use the static attribute groups assigned to the
new device.  This also fixes the possible races, too.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

It is not necessary to call device_remove_groups() when device_add_groups()
fails.

The group added by device_add_groups() should be removed if sysfs_create_link()
fails.

Fixes: fa6fdb33 ("driver core: bus_type: add dev_groups")
Signed-off-by: NJunjie Mao <junjie_mao@yeah.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

There are series of comparisons of the 'ret' variable on the failure path of
really_probe(),  so the  *switch* statement  seems  more appropriate there.
Signed-off-by: NSergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

pm_genpd_remove_device() tries hard to validate the generic PM domain
passed to it, but the validation is not complete.

dev->pm_domain contains a struct dev_pm_domain, which is the "base
class" of generic PM domains.  Other users of dev_pm_domains include
stuff like vga_switheroo.  Hence, a device could have a generic PM
domain or a vga_switcheroo PM domain in dev->pm_domain.

We need ot be certain that the PM domain is actually valid before we
try to remove it.  We can do this easily as we have a way to get the
current validated generic PM domain for a struct device.  This must
match the generic PM domain being requested for removal.

Convert the code to use this alternative validation method instead.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Acked-by: NKevin Hilman <khilman@linaro.org>
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NRafael 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: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

PM domains are rather noisy; scheduling behaviour can cause callbacks
to take longer, which causes them to spit out a warning-level message
each time a callback takes a little longer than the previous time.
There really isn't a need for this, except when debugging.
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Acked-by: NKevin Hilman <khilman@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Buses which currently supports attaching devices to their PM domains,
will invoke the dev_pm_domain_attach() API from their ->probe()
callbacks. During the attach procedure, genpd power up the PM domain.

In those scenarios where the bus/driver don't need to access its device
during probe, it may leave it in runtime PM suspended state since
that's also the default state. In that way, no notifications through
the runtime PM callbacks will reach the PM domain during probe.

For genpd, the consequence from the above scenario means the PM domain
will remain powered. Therefore, implement the struct dev_pm_domain's
->sync() callback, which is invoked from driver core after the
bus/driver has probed the device. It allows genpd to power off the PM
domain if it's unused.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>
[ Ulf: Updated patch according to updates in driver core ]
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>
Acked-by: NKevin Hilman <khilman@linaro.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

If PM domains are in use, it may be necessary to prepare the code
handling a PM domain for driver probing.  For example, in some
cases device drivers rely on the ability to power on the devices
with the help of the IO runtime PM framework and the PM domain
code needs to be ready for that.  Also, if that code has not been
fully initialized yet, the driver probing should be deferred.

Moreover, after the probing is complete, it may be necessary to
put the PM domain in question into the state reflecting the current
needs of the devices in it, for example, so that power is not drawn
in vain.  The same should be done after removing a driver from
a device, as the PM domain state may need to be changed to reflect
the new situation.

For these reasons, introduce new PM domain callbacks, ->activate,
->sync and ->dismiss called, respectively, before probing for a
device driver, after the probing has completed successfully and
if the probing has failed or the driver has been removed.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Reviewed-by: NKevin Hilman <khilman@linaro.org>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>