GitHub/moto-9609/android_kernel_motorola_exynos9610.git
7 years agoplatform/x86: intel_scu_ipc: Introduce intel_scu_ipc_raw_command()
Andy Shevchenko [Wed, 5 Apr 2017 16:05:28 +0000 (19:05 +0300)]
platform/x86: intel_scu_ipc: Introduce intel_scu_ipc_raw_command()

A new call to SCU intel_scu_ipc_raw_command() writes SPTR and DPTR
registers before sending a command.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: intel_scu_ipc: Introduce SCU_DEVICE() macro
Andy Shevchenko [Wed, 5 Apr 2017 16:05:27 +0000 (19:05 +0300)]
platform/x86: intel_scu_ipc: Introduce SCU_DEVICE() macro

For better maintainability and readability introduce a macro
for device ID table.

No functional change intended.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: intel_scu_ipc: Remove redundant subarch check
Andy Shevchenko [Wed, 5 Apr 2017 16:05:26 +0000 (19:05 +0300)]
platform/x86: intel_scu_ipc: Remove redundant subarch check

The driver is bound to the devices based on their PCI IDs.

There is no need to do an additional check.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: intel_scu_ipc: Rearrange init sequence
Andy Shevchenko [Wed, 5 Apr 2017 16:05:25 +0000 (19:05 +0300)]
platform/x86: intel_scu_ipc: Rearrange init sequence

Device pointer is used as a flag that everything is prepared.
Nevertheless the assignment happened quite before and there is a window
when a caller can get weird results or even crashes since not all fields
are initialized yet.

Rearrange initialization sequence in ->probe() to prepare everything
before use.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: intel_scu_ipc: Platform data is mandatory
Andy Shevchenko [Wed, 5 Apr 2017 16:05:24 +0000 (19:05 +0300)]
platform/x86: intel_scu_ipc: Platform data is mandatory

Fail ->probe() if there is no platform data supplied.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: silead_dmi - abort early if DMI does not match
Dmitry Torokhov [Tue, 4 Apr 2017 03:25:14 +0000 (20:25 -0700)]
platform/x86: silead_dmi - abort early if DMI does not match

There is no point in registering I2C bus notifier if DMI data does not
match.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
[andy: updated due to previous patch changed]
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: silead_dmi - do not treat all devices as i2c_clients
Dmitry Torokhov [Tue, 4 Apr 2017 03:25:13 +0000 (20:25 -0700)]
platform/x86: silead_dmi - do not treat all devices as i2c_clients

I2C bus has both i2c clients and adapter devices, so we must be careful in
notifier code and verify that we are actually dealing with an i2c client
before using it as such.

Fixes: cef9dd85acd7 ("platform/x86: add support for devices with Silead...")
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
[andy: simplified silead_ts_dmi_add_props() change]
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: asus-nb-wmi: Add wapf4 quirk for the X302UA
Santeri Toivonen [Tue, 4 Apr 2017 18:09:00 +0000 (21:09 +0300)]
platform/x86: asus-nb-wmi: Add wapf4 quirk for the X302UA

Asus laptop X302UA starts up with Wi-Fi disabled,
without a way to enable it. Set wapf=4 to fix the problem.

Signed-off-by: Santeri Toivonen <santeri.toivonen@vatsul.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: rename call_fext_func() arguments
Michał Kępień [Mon, 3 Apr 2017 09:38:59 +0000 (11:38 +0200)]
platform/x86: fujitsu-laptop: rename call_fext_func() arguments

Rename call_fext_func() arguments so that each argument's name signifies
its role:

  - cmd -> func: sub-function to call (flags, buttons etc.),
  - arg0 -> op: operation to perform (get, set, get capabilities etc.),
  - arg1 -> feature: feature to act on (e.g. which LED), if relevant,
  - arg2 -> state: state to set (e.g. LED on or off), if relevant.

Adjust whitespace to make checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: simplify call_fext_func()
Michał Kępień [Mon, 3 Apr 2017 09:38:58 +0000 (11:38 +0200)]
platform/x86: fujitsu-laptop: simplify call_fext_func()

acpi_evaluate_integer() takes a pathname parameter which contains the
name of the entity to evaluate underneath the given handle, so calling
acpi_get_handle() beforehand is redundant.  Replace the call to
acpi_get_handle() with a call to acpi_evaluate_integer(), thus
eliminating the need for a local variable storing the handle.  Adjust
whitespace to make checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: clean up local variables in call_fext_func()
Michał Kępień [Mon, 3 Apr 2017 09:38:57 +0000 (11:38 +0200)]
platform/x86: fujitsu-laptop: clean up local variables in call_fext_func()

Set values of FUNC call parameters in a designated initializer.  Do not
initialize status and handle variables as the values these are
initialized to have no influence on execution flow.  Use an array
variable instead of the address of the first element of that array.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: remove keycode fields from struct fujitsu_bl
Michał Kępień [Mon, 20 Mar 2017 09:32:24 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: remove keycode fields from struct fujitsu_bl

Remove the keycode[1-5] fields from struct fujitsu_bl as they are not
needed any more as a result of the sparse keymap migration.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: model-dependent sparse keymap overrides
Michał Kępień [Mon, 20 Mar 2017 09:32:23 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: model-dependent sparse keymap overrides

Some laptop models need to have different keycodes assigned to hotkey
scancodes.  Change the sparse keymap upon a DMI match, before the hotkey
input device is setup.

Instead of using three different callbacks in the DMI match table,
simplify code by using the driver_data field of struct dmi_system_id to
supply the requested keymap to a common callback.  Also merge keymaps
for S6410 and S6420 as they are identical.

Rename fujitsu_dmi_table to fujitsu_laptop_dmi_table to emphasize it is
no longer used by the backlight part of fujitsu-laptop.  Adjust
whitespace to make checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: use a sparse keymap for hotkey event generation
Michał Kępień [Mon, 20 Mar 2017 09:32:22 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: use a sparse keymap for hotkey event generation

Simplify hotkey event generation by using a sparse keymap.  As sparse
keymap operates on scancodes instead of keycodes, adjust variable names
and debug messages accordingly.

This patch only handles the default keymap for clarity.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
[dvhart: correct flag passed to call_fext_func in acpi_fujitsu_laptop_notify]
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: switch to a managed hotkey input device
Michał Kępień [Mon, 20 Mar 2017 09:32:21 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: switch to a managed hotkey input device

Use a managed input device for hotkey events in order to simplify two
error paths and one cleanup function while also reducing the number of
local variables required.  Remove double assignment to make checkpatch
happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: refactor hotkey input device setup
Michał Kępień [Mon, 20 Mar 2017 09:32:20 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: refactor hotkey input device setup

Simplify error handling in acpi_fujitsu_laptop_add() by moving code
responsible for setting up the input device to a separate function.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: use a sparse keymap for brightness key events
Michał Kępień [Mon, 20 Mar 2017 09:32:19 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: use a sparse keymap for brightness key events

Simplify brightness key event generation by using a sparse keymap.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: switch to a managed backlight input device
Michał Kępień [Mon, 20 Mar 2017 09:32:18 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: switch to a managed backlight input device

Use a managed input device for brightness events in order to simplify
two error paths and one cleanup function while also reducing the number
of local variables required.  Remove double assignment to make
checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: refactor backlight input device setup
Michał Kępień [Mon, 20 Mar 2017 09:32:17 +0000 (10:32 +0100)]
platform/x86: fujitsu-laptop: refactor backlight input device setup

Simplify error handling in acpi_fujitsu_bl_add() by moving code
responsible for setting up the input device to a separate function.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoMerge branch 'linux-leds/dell-laptop-changes-for-4.12'
Darren Hart [Wed, 29 Mar 2017 19:19:42 +0000 (12:19 -0700)]
Merge branch 'linux-leds/dell-laptop-changes-for-4.12'

Merge branch 'dell-laptop-changes-for-4.12' of
git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds.git
to avoid linux-next merge conflict with dell-laptop.c.

Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: silead_dmi: Add entry for Insyde 7W tablets
Hans de Goede [Wed, 22 Mar 2017 15:55:54 +0000 (16:55 +0100)]
platform/x86: silead_dmi: Add entry for Insyde 7W tablets

Add an entry providing the necessary info to make the touchscreen work
on various tablets based on the Insyde 7W whitebox tablet.

This has been tested on a DEXP Ursus 7W tablet.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: silead_dmi: Constify properties arrays
Hans de Goede [Wed, 22 Mar 2017 15:55:53 +0000 (16:55 +0100)]
platform/x86: silead_dmi: Constify properties arrays

Now that device_add_properties takes a const property_entry * rather
then a non-const one we can constify the properties arrays.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: dell-*: Call new led hw_changed API on kbd brightness change
Hans de Goede [Thu, 16 Mar 2017 10:55:35 +0000 (11:55 +0100)]
platform/x86: dell-*: Call new led hw_changed API on kbd brightness change

Make dell-wmi notify on hotkey kbd brightness changes, listen for this
in dell-laptop and call led_classdev_notify_brightness_hw_changed.

This will allow userspace to monitor (poll) for brightness changes on
these LEDs caused by the hotkey.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: dell-laptop: Protect kbd_state against races
Hans de Goede [Thu, 16 Mar 2017 10:55:34 +0000 (11:55 +0100)]
platform/x86: dell-laptop: Protect kbd_state against races

The kbd led code has multiple entry points each of which modifies the
kbd_state by reading it, modifying a copy, writing the copy and on
error setting the modified copy writing back the original state.

This is racy, so add a mutex protection the read-modify-write cycle
on each of the entry points.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: dell-laptop: Refactor kbd_led_triggers_store()
Hans de Goede [Thu, 16 Mar 2017 10:55:33 +0000 (11:55 +0100)]
platform/x86: dell-laptop: Refactor kbd_led_triggers_store()

Return -EINVAL immediately on invalid input, rather then doing
the straight path in an if block and returning -EINVAL at the end
of the function.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: dell-*: Add a generic dell-laptop notifier chain
Hans de Goede [Thu, 16 Mar 2017 10:55:32 +0000 (11:55 +0100)]
platform/x86: dell-*: Add a generic dell-laptop notifier chain

There are several cases where events handled in one of the dell-* drivers
need to be propagated to another dell-* driver.

This commit adds 3 generic functions:
dell_laptop_register_notifier()
dell_laptop_unregister_notifier()
dell_laptop_call_notifier()

It currently only defines 1 action:
DELL_LAPTOP_KBD_BACKLIGHT_BRIGHTNESS_CHANGED

Which is intended to propagate kbd_backlight_brightness_changed wmi
events from dell-wmi to dell-laptop (which contains the actual kbd
backlight driver).

These functions are put in dell-smbios as both dell-wmi and dell-laptop
use smbios functions and I do not want to put the notifier head in
either driver, as that will make the 2 drivers depend on each other.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: remove pf_device field from struct fujitsu_bl
Michał Kępień [Tue, 14 Mar 2017 10:26:31 +0000 (11:26 +0100)]
platform/x86: fujitsu-laptop: remove pf_device field from struct fujitsu_bl

Both struct fujitsu_bl and struct fujitsu_laptop have a pf_device field.
Up until now the latter has been redundant, which is logically incorrect
because the primary function of the platform device created by
fujitsu-laptop is to provide laptop-related (not brightness-related)
attributes to userspace.

Remove the pf_device field from struct fujitsu_bl and make all module
code use the pf_device field of struct fujitsu_laptop instead.

Suggested-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Tested-by: Jonathan Woithe <jwoithe@just42.net>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: only register platform device if FUJ02E3 is present
Michał Kępień [Tue, 14 Mar 2017 10:26:30 +0000 (11:26 +0100)]
platform/x86: fujitsu-laptop: only register platform device if FUJ02E3 is present

The platform device registered by fujitsu-laptop is registered
unconditionally while sysfs attributes attached to it depend on the
FUJ02E3 ACPI device being present.  Fix this by moving platform device
creation and removal to acpi_fujitsu_laptop_add() and
acpi_fujitsu_laptop_remove(), respectively.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Tested-by: Jonathan Woithe <jwoithe@just42.net>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: add and remove platform device in separate functions
Michał Kępień [Tue, 14 Mar 2017 10:26:29 +0000 (11:26 +0100)]
platform/x86: fujitsu-laptop: add and remove platform device in separate functions

Platform device handling adds a lot of complexity to fujitsu_init(),
which hinders its readability.  Extract code responsible for adding and
removing the platform device to separate functions.  Adjust whitespace
to make checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Tested-by: Jonathan Woithe <jwoithe@just42.net>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: simplify platform device attribute definitions
Michał Kępień [Tue, 14 Mar 2017 10:26:28 +0000 (11:26 +0100)]
platform/x86: fujitsu-laptop: simplify platform device attribute definitions

Use the DEVICE_ATTR_RO() macro to get rid of ignore_store() and shorten
attribute definitions.  Adjust whitespace to make checkpatch happy.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Tested-by: Jonathan Woithe <jwoithe@just42.net>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: fujitsu-laptop: remove backlight-related attributes from the platform...
Michał Kępień [Tue, 14 Mar 2017 10:26:27 +0000 (11:26 +0100)]
platform/x86: fujitsu-laptop: remove backlight-related attributes from the platform device

Setting backlight level using a vendor-specific interface should only be
possible when using the latter is either explicitly requested by the
user or automatically selected by the kernel.  fujitsu-laptop violates
that premise by unconditionally attaching three backlight-related
attributes to the platform device it registers.  Remove the offending
attributes from the platform device.  Update module comments to reflect
this change.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Tested-by: Jonathan Woithe <jwoithe@just42.net>
Reviewed-by: Jonathan Woithe <jwoithe@just42.net>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: eeepc-laptop: Skip unknown key messages 0x50 0x51
Pau Espin Pedrol [Fri, 10 Feb 2017 17:48:37 +0000 (18:48 +0100)]
platform/x86: eeepc-laptop: Skip unknown key messages 0x50 0x51

Otherwise those are printed several times when (un)plugging the AC
connector.

Signed-off-by: Pau Espin Pedrol <pespin.shar@gmail.com>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
7 years agoplatform/x86: intel-hid: do not set parents of input devices explicitly
Michał Kępień [Thu, 9 Mar 2017 09:36:51 +0000 (10:36 +0100)]
platform/x86: intel-hid: do not set parents of input devices explicitly

devm_input_allocate_device() already causes the supplied struct device
to be set as the parent of the input device, so doing it again is
redundant.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Alex Hung <alex.hung@canonical.com>
7 years agoplatform/x86: toshiba_acpi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:50 +0000 (13:11 +0100)]
platform/x86: toshiba_acpi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: toshiba-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:49 +0000 (13:11 +0100)]
platform/x86: toshiba-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: topstar-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:48 +0000 (13:11 +0100)]
platform/x86: topstar-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: panasonic-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:47 +0000 (13:11 +0100)]
platform/x86: panasonic-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

As this reduces acpi_pcc_destroy_input() to one line, replace all calls
to that function with direct calls to input_unregister_device().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: msi-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:46 +0000 (13:11 +0100)]
platform/x86: msi-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: msi-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:45 +0000 (13:11 +0100)]
platform/x86: msi-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

As this reduces msi_laptop_input_destroy() to one line, replace all
calls to that function with direct calls to input_unregister_device().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: ideapad-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:44 +0000 (13:11 +0100)]
platform/x86: ideapad-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: hp-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:43 +0000 (13:11 +0100)]
platform/x86: hp-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: eeepc-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:42 +0000 (13:11 +0100)]
platform/x86: eeepc-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: dell-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:41 +0000 (13:11 +0100)]
platform/x86: dell-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

As this reduces dell_wmi_input_destroy() to one line, replace all calls
to that function with direct calls to input_unregister_device().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: dell-wmi-aio: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:40 +0000 (13:11 +0100)]
platform/x86: dell-wmi-aio: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: asus-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:39 +0000 (13:11 +0100)]
platform/x86: asus-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: asus-laptop: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:38 +0000 (13:11 +0100)]
platform/x86: asus-laptop: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: acer-wmi: remove sparse_keymap_free() calls
Michał Kępień [Thu, 9 Mar 2017 12:11:37 +0000 (13:11 +0100)]
platform/x86: acer-wmi: remove sparse_keymap_free() calls

As sparse_keymap_setup() now uses a managed memory allocation for the
keymap copy it creates, the latter is freed automatically.  Remove all
calls to sparse_keymap_free().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoMerge branch 'ib/4.10-sparse-keymap-managed' of git://git.kernel.org/pub/scm/linux...
Darren Hart [Wed, 15 Mar 2017 05:57:36 +0000 (22:57 -0700)]
Merge branch 'ib/4.10-sparse-keymap-managed' of git://git./linux/kernel/git/dtor/input into testing

7 years agoplatform/x86: fujitsu-laptop: cleanup error labels in fujitsu_init()
Michał Kępień [Fri, 10 Mar 2017 10:50:36 +0000 (11:50 +0100)]
platform/x86: fujitsu-laptop: cleanup error labels in fujitsu_init()

Error labels currently used in fujitsu_init() are really hard to follow:
some (fail_laptop) indicate which operation has failed, others
(fail_sysfs_group) indicate where unrolling should start and the rest
(fail_platform_driver) is simply confusing.  Change them to follow the
pattern used throughout the rest of the module, i.e. make every label
indicate the first unrolling operation it leads to.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: fujitsu-laptop: only register backlight device if FUJ02B1 is present
Michał Kępień [Fri, 10 Mar 2017 10:50:35 +0000 (11:50 +0100)]
platform/x86: fujitsu-laptop: only register backlight device if FUJ02B1 is present

As the backlight device registered by fujitsu-laptop relies on the
FUJ02B1 ACPI device being present, only register the backlight device
once that ACPI device is detected.

Suggested-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: fujitsu-laptop: sync backlight power status in acpi_fujitsu_laptop_add()
Michał Kępień [Fri, 10 Mar 2017 10:50:34 +0000 (11:50 +0100)]
platform/x86: fujitsu-laptop: sync backlight power status in acpi_fujitsu_laptop_add()

Registering an ACPI driver does not mean the device it handles has to
exist.  As the code which syncs backlight power status uses
call_fext_func(), it needs the FUJ02E3 ACPI device to be present, so
ensure that code is only run once the FUJ02E3 device is detected.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: fujitsu-laptop: register backlight device in a separate function
Michał Kępień [Fri, 10 Mar 2017 10:50:33 +0000 (11:50 +0100)]
platform/x86: fujitsu-laptop: register backlight device in a separate function

Move code responsible for backlight device registration to a separate
function in order to simplify error handling and decrease indentation.
Simplify initialization of struct backlight_properties.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: fujitsu-laptop: simplify brightness key event generation logic
Michał Kępień [Wed, 1 Mar 2017 06:42:54 +0000 (07:42 +0100)]
platform/x86: fujitsu-laptop: simplify brightness key event generation logic

Returning early when there is no brightness change allows removal of a
duplicate code block, makes the purpose of the following code clearer
and allows the condition surrounding key event generation to be removed.
Local integer variables can also be declared in a single line.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: fujitsu-laptop: decrease indentation in acpi_fujitsu_bl_notify()
Michał Kępień [Wed, 1 Mar 2017 06:42:53 +0000 (07:42 +0100)]
platform/x86: fujitsu-laptop: decrease indentation in acpi_fujitsu_bl_notify()

acpi_fujitsu_bl_notify() is pretty deeply nested, which hurts
readability.  Strip off one level of indentation by returning early when
the event code supplied as argument is not ACPI_FUJITSU_NOTIFY_CODE1.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Jonathan Woithe <jwoithe@just42.net>
7 years agoplatform/x86: ideapad-laptop: Add IdeaPad 310-15IKB to no_hw_rfkill
Sven Rebhan [Tue, 21 Feb 2017 19:53:48 +0000 (20:53 +0100)]
platform/x86: ideapad-laptop: Add IdeaPad 310-15IKB to no_hw_rfkill

Like other Lenovo models the IdeaPad 310-15IKB does not have an hw rfkill
switch. This results in hard-blocked radios after boot, resulting in
always blocked radios rendering them unusable.

Add the IdeaPad 310-15IKB to the no_hw_rfkill DMI list and allows using
the built-in radios.

Signed-off-by: Sven Rebhan <Sven.Rebhan@googlemail.com>
[andy: massaged commit message]
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: thinkpad_acpi: add mapping for new hotkeys
Christian Kellner [Tue, 28 Feb 2017 16:10:57 +0000 (17:10 +0100)]
platform/x86: thinkpad_acpi: add mapping for new hotkeys

The T470, X270 emits new hkey events in the 0x1311 - 0x1315 range.
According to the user manual they should launch a user selected
favorite application (star icon, 0x1311), snipping tool (0x1312,
currently ignored), enable/disable bluetooth (0x1314) and open they
keyboard settings (0x1315).

The third nibble (0xf00) is used to differentiate between the original
hotkeys, the adaptive keyboard codes and the new, additional ones.

Signed-off-by: Christian Kellner <ckellner@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: thinkpad_acpi: guard generic hotkey case
Christian Kellner [Tue, 28 Feb 2017 16:10:56 +0000 (17:10 +0100)]
platform/x86: thinkpad_acpi: guard generic hotkey case

Currently when dispatching hotkeys we check if the scancode is in
the range of 0 and TPACPI_HOTKEY_MAP_LEN, although the bottom 20
entries in the hotkey keymap are already adaptive keycodes.
Therefore we introduce a TP_ACPI_HOTKEYSCAN_ADAPTIVE_START and
ensure that we are in the range 0 and ADAPTIVE_START for the generic
keycode case.

Signed-off-by: Christian Kellner <ckellner@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
7 years agoplatform/x86: intel-hid: remove redundant set_bit() call
Michał Kępień [Fri, 24 Feb 2017 10:33:09 +0000 (11:33 +0100)]
platform/x86: intel-hid: remove redundant set_bit() call

Remove redundant set_bit() call for KEY_RFKILL as it is already made by
sparse_keymap_setup() due to KEY_RFKILL being assigned to event code 8
in intel_hid_keymap.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Alex Hung <alex.hung@canonical.com>
7 years agoplatform/x86: intel-hid: use devm_input_allocate_device() for HID events input device
Michał Kępień [Fri, 24 Feb 2017 10:33:08 +0000 (11:33 +0100)]
platform/x86: intel-hid: use devm_input_allocate_device() for HID events input device

intel_hid_input_setup() is only called from the platform driver's .probe
callback.  Use the devm variant of input_allocate_device() for
allocating memory for the HID events input device in order to simplify
two error paths and get rid of intel_hid_input_destroy().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Alex Hung <alex.hung@canonical.com>
7 years agoplatform/x86: intel-hid: make intel_hid_set_enable() take a boolean argument
Michał Kępień [Fri, 24 Feb 2017 10:33:07 +0000 (11:33 +0100)]
platform/x86: intel-hid: make intel_hid_set_enable() take a boolean argument

As the integer value passed to intel_hid_set_enable() is always
explicitly passed and is used solely as a boolean value, make it a bool.

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Alex Hung <alex.hung@canonical.com>
7 years agoplatform/x86: intel-hid: simplify enabling/disabling HID events
Michał Kępień [Fri, 24 Feb 2017 10:33:06 +0000 (11:33 +0100)]
platform/x86: intel-hid: simplify enabling/disabling HID events

ACPI method HDSM takes a single integer argument.  Use
acpi_execute_simple_method() instead of acpi_evaluate_object() for
calling that ACPI method to simplify code and reduce the number of local
variables inside intel_hid_set_enable().

Signed-off-by: Michał Kępień <kernel@kempniu.pl>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-and-tested-by: Alex Hung <alex.hung@canonical.com>
7 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Wed, 15 Mar 2017 04:31:23 +0000 (21:31 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Ensure that mtu is at least IPV6_MIN_MTU in ipv6 VTI tunnel driver,
    from Steffen Klassert.

 2) Fix crashes when user tries to get_next_key on an LPM bpf map, from
    Alexei Starovoitov.

 3) Fix detection of VLAN fitlering feature for bnx2x VF devices, from
    Michal Schmidt.

 4) We can get a divide by zero when TCP socket are morphed into
    listening state, fix from Eric Dumazet.

 5) Fix socket refcounting bugs in skb_complete_wifi_ack() and
    skb_complete_tx_timestamp(). From Eric Dumazet.

 6) Use after free in dccp_feat_activate_values(), also from Eric
    Dumazet.

 7) Like bonding team needs to use ETH_MAX_MTU as netdev->max_mtu, from
    Jarod Wilson.

 8) Fix use after free in vrf_xmit(), from David Ahern.

 9) Don't do UDP Fragmentation Offload on IPComp ipsec packets, from
    Alexey Kodanev.

10) Properly check napi_complete_done() return value in order to decide
    whether to re-enable IRQs or not in amd-xgbe driver, from Thomas
    Lendacky.

11) Fix double free of hwmon device in marvell phy driver, from Andrew
    Lunn.

12) Don't crash on malformed netlink attributes in act_connmark, from
    Etienne Noss.

13) Don't remove routes with a higher metric in ipv6 ECMP route replace,
    from Sabrina Dubroca.

14) Don't write into a cloned SKB in ipv6 fragmentation handling, from
    Florian Westphal.

15) Fix routing redirect races in dccp and tcp, basically the ICMP
    handler can't modify the socket's cached route in it's locked by the
    user at this moment. From Jon Maxwell.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (108 commits)
  qed: Enable iSCSI Out-of-Order
  qed: Correct out-of-bound access in OOO history
  qed: Fix interrupt flags on Rx LL2
  qed: Free previous connections when releasing iSCSI
  qed: Fix mapping leak on LL2 rx flow
  qed: Prevent creation of too-big u32-chains
  qed: Align CIDs according to DORQ requirement
  mlxsw: reg: Fix SPVMLR max record count
  mlxsw: reg: Fix SPVM max record count
  net: Resend IGMP memberships upon peer notification.
  dccp: fix memory leak during tear-down of unsuccessful connection request
  tun: fix premature POLLOUT notification on tun devices
  dccp/tcp: fix routing redirect race
  ucc/hdlc: fix two little issue
  vxlan: fix ovs support
  net: use net->count to check whether a netns is alive or not
  bridge: drop netfilter fake rtable unconditionally
  ipv6: avoid write to a possibly cloned skb
  net: wimax/i2400m: fix NULL-deref at probe
  isdn/gigaset: fix NULL-deref at probe
  ...

7 years agoMerge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Tue, 14 Mar 2017 22:11:19 +0000 (15:11 -0700)]
Merge branch 'for-4.11-fixes' of git://git./linux/kernel/git/tj/cgroup

Pull cgroup fixes from Tejun Heo:
 "Three cgroup fixes.  Nothing critical:

   - the pids controller could trigger suspicious RCU warning
     spuriously. Fixed.

   - in the debug controller, %p -> %pK to protect kernel pointer
     from getting exposed.

   - documentation formatting fix"

* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroups: censor kernel pointer in debug files
  cgroup/pids: remove spurious suspicious RCU usage warning
  cgroup: Fix indenting in PID controller documentation

7 years agoMerge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Tue, 14 Mar 2017 22:00:43 +0000 (15:00 -0700)]
Merge branch 'for-4.11-fixes' of git://git./linux/kernel/git/tj/libata

Pull libata fixes from Tejun Heo:
 "Three libata fixes:

   - fix for a circular reference bug in sysfs code which prevented
     pata_legacy devices from being released after probe failure, which
     in turn prevented devres from releasing the associated resources.

   - drop spurious WARN in the command issue path which can be triggered
     by a legitimate passthrough command.

   - an ahci_qoriq specific fix"

* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
  ahci: qoriq: correct the sata ecc setting error
  libata: drop WARN from protocol error in ata_sff_qc_issue()
  libata: transport: Remove circular dependency at free time

7 years agoMerge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Linus Torvalds [Tue, 14 Mar 2017 21:52:08 +0000 (14:52 -0700)]
Merge branch 'for-4.11-fixes' of git://git./linux/kernel/git/tj/wq

Pull workqueue fix from Tejun Heo:
 "If a delayed work is queued with NULL @wq, workqueue code explodes
  after the timer expires at which point it's difficult to tell who the
  culprit was.

  This actually happened and the offender was net/smc this time.

  Add an explicit sanity check for it in the queueing path"

* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: trigger WARN if queue_delayed_work() is called with NULL @wq

7 years agoMerge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Tue, 14 Mar 2017 21:48:50 +0000 (14:48 -0700)]
Merge branch 'for-4.11-fixes' of git://git./linux/kernel/git/tj/percpu

Pull percpu fixes from Tejun Heo:

 - the allocation path was updating pcpu_nr_empty_pop_pages without the
   required locking which can lead to incorrect handling of empty chunks
   (e.g. keeping too many around), which is buggy but shouldn't lead to
   critical failures. Fixed by adding the locking

 - a trivial patch to drop an unused param from pcpu_get_pages()

* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
  percpu: remove unused chunk_alloc parameter from pcpu_get_pages()
  percpu: acquire pcpu_lock when updating pcpu_nr_empty_pop_pages

7 years agoMerge branch 'qed-fixes'
David S. Miller [Tue, 14 Mar 2017 18:37:06 +0000 (11:37 -0700)]
Merge branch 'qed-fixes'

Yuval Mintz says:

====================
qed: Fixes series

This address several different issues in qed.
The more significant portions:

Patch #1 would cause timeout when qedr utilizes the highest
CIDs availble for it [or when future qede adapters would utilize
queues in some constellations].

Patch #4 fixes a leak of mapped addresses; When iommu is enabled,
offloaded storage protocols might eventually run out of resources
and fail to map additional buffers.

Patches #6,#7 were missing in the initial iSCSI infrastructure
submissions, and would hamper qedi's stability when it reaches
out-of-order scenarios.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Enable iSCSI Out-of-Order
Mintz, Yuval [Tue, 14 Mar 2017 13:26:04 +0000 (15:26 +0200)]
qed: Enable iSCSI Out-of-Order

Missing in the initial submission, qed fails to propagate qedi's
request to enable OOO to firmware.

Fixes: fc831825f99e ("qed: Add support for hardware offloaded iSCSI")
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Correct out-of-bound access in OOO history
Mintz, Yuval [Tue, 14 Mar 2017 13:26:03 +0000 (15:26 +0200)]
qed: Correct out-of-bound access in OOO history

Need to set the number of entries in database, otherwise the logic
would quickly surpass the array.

Fixes: 1d6cff4fca43 ("qed: Add iSCSI out of order packet handling")
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Fix interrupt flags on Rx LL2
Ram Amrani [Tue, 14 Mar 2017 13:26:02 +0000 (15:26 +0200)]
qed: Fix interrupt flags on Rx LL2

Before iterating over the the LL2 Rx ring, the ring's
spinlock is taken via spin_lock_irqsave().
The actual processing of the packet [including handling
by the protocol driver] is done without said lock,
so qed releases the spinlock and re-claims it afterwards.

Problem is that the final spin_lock_irqrestore() at the end
of the iteration uses the original flags saved from the
initial irqsave() instead of the flags from the most recent
irqsave(). So it's possible that the interrupt status would
be incorrect at the end of the processing.

Fixes: 0a7fb11c23c0 ("qed: Add Light L2 support");
CC: Ram Amrani <Ram.Amrani@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Free previous connections when releasing iSCSI
Mintz, Yuval [Tue, 14 Mar 2017 13:26:01 +0000 (15:26 +0200)]
qed: Free previous connections when releasing iSCSI

Fixes: fc831825f99e ("qed: Add support for hardware offloaded iSCSI")
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Fix mapping leak on LL2 rx flow
Mintz, Yuval [Tue, 14 Mar 2017 13:26:00 +0000 (15:26 +0200)]
qed: Fix mapping leak on LL2 rx flow

When receiving an Rx LL2 packet, qed fails to unmap the previous buffer.

Fixes: 0a7fb11c23c0 ("qed: Add Light L2 support");
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Prevent creation of too-big u32-chains
Tomer Tayar [Tue, 14 Mar 2017 13:25:59 +0000 (15:25 +0200)]
qed: Prevent creation of too-big u32-chains

Current Logic would allow the creation of a chain with U32_MAX + 1
elements, when the actual maximum supported by the driver infrastructure
is U32_MAX.

Fixes: a91eb52abb50 ("qed: Revisit chain implementation")
Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Align CIDs according to DORQ requirement
Ram Amrani [Tue, 14 Mar 2017 13:25:58 +0000 (15:25 +0200)]
qed: Align CIDs according to DORQ requirement

The Doorbell HW block can be configured at a granularity
of 16 x CIDs, so we need to make sure that the actual number
of CIDs configured would be a multiplication of 16.

Today, when RoCE is enabled - given that the number is unaligned,
doorbelling the higher CIDs would fail to reach the firmware and
would eventually timeout.

Fixes: dbb799c39717 ("qed: Initialize hardware for new protocols")
Signed-off-by: Ram Amrani <Ram.Amrani@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'mlxsw-small-fixes'
David S. Miller [Tue, 14 Mar 2017 18:35:11 +0000 (11:35 -0700)]
Merge branch 'mlxsw-small-fixes'

Jiri Pirko says:

====================
mlxsw: Couple of fixes

Couple or small fixes.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: reg: Fix SPVMLR max record count
Jiri Pirko [Tue, 14 Mar 2017 13:00:01 +0000 (14:00 +0100)]
mlxsw: reg: Fix SPVMLR max record count

The num_rec field is 8 bit, so the maximal count number is 255.
This fixes vlans learning not being enabled for wider ranges than 255.

Fixes: a4feea74cd7a ("mlxsw: reg: Add Switch Port VLAN MAC Learning register definition")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: reg: Fix SPVM max record count
Jiri Pirko [Tue, 14 Mar 2017 13:00:00 +0000 (14:00 +0100)]
mlxsw: reg: Fix SPVM max record count

The num_rec field is 8 bit, so the maximal count number is 255. This
fixes vlans not being enabled for wider ranges than 255.

Fixes: b2e345f9a454 ("mlxsw: reg: Add Switch Port VID and Switch Port VLAN Membership registers definitions")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: Resend IGMP memberships upon peer notification.
Vlad Yasevich [Tue, 14 Mar 2017 12:58:08 +0000 (08:58 -0400)]
net: Resend IGMP memberships upon peer notification.

When we notify peers of potential changes,  it's also good to update
IGMP memberships.  For example, during VM migration, updating IGMP
memberships will redirect existing multicast streams to the VM at the
new location.

Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodccp: fix memory leak during tear-down of unsuccessful connection request
Hannes Frederic Sowa [Sun, 12 Mar 2017 23:01:30 +0000 (00:01 +0100)]
dccp: fix memory leak during tear-down of unsuccessful connection request

This patch fixes a memory leak, which happens if the connection request
is not fulfilled between parsing the DCCP options and handling the SYN
(because e.g. the backlog is full), because we forgot to free the
list of ack vectors.

Reported-by: Jianwen Ji <jiji@redhat.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotun: fix premature POLLOUT notification on tun devices
Hannes Frederic Sowa [Sun, 12 Mar 2017 23:00:26 +0000 (00:00 +0100)]
tun: fix premature POLLOUT notification on tun devices

aszlig observed failing ssh tunnels (-w) during initialization since
commit cc9da6cc4f56e0 ("ipv6: addrconf: use stable address generator for
ARPHRD_NONE"). We already had reports that the mentioned commit breaks
Juniper VPN connections. I can't clearly say that the Juniper VPN client
has the same problem, but it is worth a try to hint to this patch.

Because of the early generation of link local addresses, the kernel now
can start asking for routers on the local subnet much earlier than usual.
Those router solicitation packets arrive inside the ssh channels and
should be transmitted to the tun fd before the configuration scripts
might have upped the interface and made it ready for transmission.

ssh polls on the interface and receives back a POLL_OUT. It tries to send
the earily router solicitation packet to the tun interface.  Unfortunately
it hasn't been up'ed yet by config scripts, thus failing with -EIO. ssh
doesn't retry again and considers the tun interface broken forever.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=121131
Fixes: cc9da6cc4f56 ("ipv6: addrconf: use stable address generator for ARPHRD_NONE")
Cc: Bjørn Mork <bjorn@mork.no>
Reported-by: Valdis Kletnieks <Valdis.Kletnieks@vt.edu>
Cc: Valdis Kletnieks <Valdis.Kletnieks@vt.edu>
Reported-by: Jonas Lippuner <jonas@lippuner.ca>
Cc: Jonas Lippuner <jonas@lippuner.ca>
Reported-by: aszlig <aszlig@redmoonstudios.org>
Cc: aszlig <aszlig@redmoonstudios.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodccp/tcp: fix routing redirect race
Jon Maxwell [Fri, 10 Mar 2017 05:40:33 +0000 (16:40 +1100)]
dccp/tcp: fix routing redirect race

As Eric Dumazet pointed out this also needs to be fixed in IPv6.
v2: Contains the IPv6 tcp/Ipv6 dccp patches as well.

We have seen a few incidents lately where a dst_enty has been freed
with a dangling TCP socket reference (sk->sk_dst_cache) pointing to that
dst_entry. If the conditions/timings are right a crash then ensues when the
freed dst_entry is referenced later on. A Common crashing back trace is:

 #8 [] page_fault at ffffffff8163e648
    [exception RIP: __tcp_ack_snd_check+74]
.
.
 #9 [] tcp_rcv_established at ffffffff81580b64
#10 [] tcp_v4_do_rcv at ffffffff8158b54a
#11 [] tcp_v4_rcv at ffffffff8158cd02
#12 [] ip_local_deliver_finish at ffffffff815668f4
#13 [] ip_local_deliver at ffffffff81566bd9
#14 [] ip_rcv_finish at ffffffff8156656d
#15 [] ip_rcv at ffffffff81566f06
#16 [] __netif_receive_skb_core at ffffffff8152b3a2
#17 [] __netif_receive_skb at ffffffff8152b608
#18 [] netif_receive_skb at ffffffff8152b690
#19 [] vmxnet3_rq_rx_complete at ffffffffa015eeaf [vmxnet3]
#20 [] vmxnet3_poll_rx_only at ffffffffa015f32a [vmxnet3]
#21 [] net_rx_action at ffffffff8152bac2
#22 [] __do_softirq at ffffffff81084b4f
#23 [] call_softirq at ffffffff8164845c
#24 [] do_softirq at ffffffff81016fc5
#25 [] irq_exit at ffffffff81084ee5
#26 [] do_IRQ at ffffffff81648ff8

Of course it may happen with other NIC drivers as well.

It's found the freed dst_entry here:

 224 static bool tcp_in_quickack_mode(struct sock *sk)↩
 225 {↩
 226 ▹       const struct inet_connection_sock *icsk = inet_csk(sk);↩
 227 ▹       const struct dst_entry *dst = __sk_dst_get(sk);↩
 228 ↩
 229 ▹       return (dst && dst_metric(dst, RTAX_QUICKACK)) ||↩
 230 ▹       ▹       (icsk->icsk_ack.quick && !icsk->icsk_ack.pingpong);↩
 231 }↩

But there are other backtraces attributed to the same freed dst_entry in
netfilter code as well.

All the vmcores showed 2 significant clues:

- Remote hosts behind the default gateway had always been redirected to a
different gateway. A rtable/dst_entry will be added for that host. Making
more dst_entrys with lower reference counts. Making this more probable.

- All vmcores showed a postitive LockDroppedIcmps value, e.g:

LockDroppedIcmps                  267

A closer look at the tcp_v4_err() handler revealed that do_redirect() will run
regardless of whether user space has the socket locked. This can result in a
race condition where the same dst_entry cached in sk->sk_dst_entry can be
decremented twice for the same socket via:

do_redirect()->__sk_dst_check()-> dst_release().

Which leads to the dst_entry being prematurely freed with another socket
pointing to it via sk->sk_dst_cache and a subsequent crash.

To fix this skip do_redirect() if usespace has the socket locked. Instead let
the redirect take place later when user space does not have the socket
locked.

The dccp/IPv6 code is very similar in this respect, so fixing it there too.

As Eric Garver pointed out the following commit now invalidates routes. Which
can set the dst->obsolete flag so that ipv4_dst_check() returns null and
triggers the dst_release().

Fixes: ceb3320610d6 ("ipv4: Kill routes during PMTU/redirect updates.")
Cc: Eric Garver <egarver@redhat.com>
Cc: Hannes Sowa <hsowa@redhat.com>
Signed-off-by: Jon Maxwell <jmaxwell37@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoucc/hdlc: fix two little issue
Zhao Qiang [Tue, 14 Mar 2017 01:38:33 +0000 (09:38 +0800)]
ucc/hdlc: fix two little issue

1. modify bd_status from u32 to u16 in function hdlc_rx_done,
because bd_status register is 16bits
2. write bd_length register before writing bd_status register

Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'powerpc-4.11-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Tue, 14 Mar 2017 02:48:22 +0000 (19:48 -0700)]
Merge tag 'powerpc-4.11-4' of git://git./linux/kernel/git/powerpc/linux

Pull some more powerpc fixes from Michael Ellerman:
 "The main item is the addition of the Power9 Machine Check handler.
  This was delayed to make sure some details were correct, and is as
  minimal as possible.

  The rest is small fixes, two for the Power9 PMU, two dealing with
  obscure toolchain problems, two for the PowerNV IOMMU code (used by
  VFIO), and one to fix a crash on 32-bit machines with macio devices
  due to missing dma_ops.

  Thanks to:
    Alexey Kardashevskiy, Cyril Bur, Larry Finger, Madhavan Srinivasan,
    Nicholas Piggin"

* tag 'powerpc-4.11-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/64s: POWER9 machine check handler
  powerpc/64s: allow machine check handler to set severity and initiator
  powerpc/64s: fix handling of non-synchronous machine checks
  powerpc/pmac: Fix crash in dma-mapping.h with NULL dma_ops
  powerpc/powernv/ioda2: Update iommu table base on ownership change
  powerpc/powernv/ioda2: Gracefully fail if too many TCE levels requested
  selftests/powerpc: Replace stxvx and lxvx with stxvd2x/lxvd2x
  powerpc/perf: Handle sdar_mode for marked event in power9
  powerpc/perf: Fix perf_get_data_addr() for power9 DD1
  powerpc/boot: Fix zImage TOC alignment

7 years agovxlan: fix ovs support
Nicolas Dichtel [Mon, 13 Mar 2017 15:24:03 +0000 (16:24 +0100)]
vxlan: fix ovs support

The required changes in the function vxlan_dev_create() were missing
in commit 8bcdc4f3a20b.
The vxlan device is not registered anymore after this patch and the error
path causes an stack dump:
 WARNING: CPU: 3 PID: 1498 at net/core/dev.c:6713 rollback_registered_many+0x9d/0x3f0

Fixes: 8bcdc4f3a20b ("vxlan: add changelink support")
CC: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: use net->count to check whether a netns is alive or not
Andrey Vagin [Mon, 13 Mar 2017 04:36:18 +0000 (21:36 -0700)]
net: use net->count to check whether a netns is alive or not

The previous idea was to check whether a net namespace is in
net_exit_list or not. It doesn't work, because net->exit_list is used in
__register_pernet_operations and __unregister_pernet_operations where
all namespaces are added to a temporary list to make cleanup in a error
case, so list_empty(&net->exit_list) always returns false.

Reported-by: Mantas Mikulėnas <grawity@gmail.com>
Fixes: 002d8a1a6c11 ("net: skip genenerating uevents for network namespaces that are exiting")
Signed-off-by: Andrei Vagin <avagin@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'platform-drivers-x86-v4.11-2' of git://git.infradead.org/linux-platform...
Linus Torvalds [Mon, 13 Mar 2017 20:23:43 +0000 (13:23 -0700)]
Merge tag 'platform-drivers-x86-v4.11-2' of git://git.infradead.org/linux-platform-drivers-x86

Pull x86 platform driver updates from Darren Hart:
 "Asus fixes for the airplane LED and a long awaited fujitsu cleanup.

  asus-wmi:
   - Remove quirk_no_rfkill
   - Detect quirk_no_rfkill from the DSDT

  fujitsu-laptop:
   - remove redundant MODULE_ALIAS entries
   - autodetect LCD interface on all models
   - simplify acpi_bus_register_driver() error handling
   - remove redundant forward declarations
   - replace numeric values with constants
   - rename FUNC_RFKILL to FUNC_FLAGS
   - make platform-related variables match naming convention
   - replace "hotkey" with "laptop" in symbol names
   - clearly denote backlight-related symbols"

* tag 'platform-drivers-x86-v4.11-2' of git://git.infradead.org/linux-platform-drivers-x86:
  platform/x86: asus-wmi: Remove quirk_no_rfkill
  platform/x86: asus-wmi: Detect quirk_no_rfkill from the DSDT
  platform/x86: fujitsu-laptop: remove redundant MODULE_ALIAS entries
  platform/x86: fujitsu-laptop: autodetect LCD interface on all models
  platform/x86: fujitsu-laptop: simplify acpi_bus_register_driver() error handling
  platform/x86: fujitsu-laptop: remove redundant forward declarations
  platform/x86: fujitsu-laptop: replace numeric values with constants
  platform/x86: fujitsu-laptop: rename FUNC_RFKILL to FUNC_FLAGS
  platform/x86: fujitsu-laptop: make platform-related variables match naming convention
  platform/x86: fujitsu-laptop: replace "hotkey" with "laptop" in symbol names
  platform/x86: fujitsu-laptop: clearly denote backlight-related symbols

7 years agobridge: drop netfilter fake rtable unconditionally
Florian Westphal [Mon, 13 Mar 2017 16:38:17 +0000 (17:38 +0100)]
bridge: drop netfilter fake rtable unconditionally

Andreas reports kernel oops during rmmod of the br_netfilter module.
Hannes debugged the oops down to a NULL rt6info->rt6i_indev.

Problem is that br_netfilter has the nasty concept of adding a fake
rtable to skb->dst; this happens in a br_netfilter prerouting hook.

A second hook (in bridge LOCAL_IN) is supposed to remove these again
before the skb is handed up the stack.

However, on module unload hooks get unregistered which means an
skb could traverse the prerouting hook that attaches the fake_rtable,
while the 'fake rtable remove' hook gets removed from the hooklist
immediately after.

Fixes: 34666d467cbf1e2e3c7 ("netfilter: bridge: move br_netfilter out of the core")
Reported-by: Andreas Karis <akaris@redhat.com>
Debugged-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipv6: avoid write to a possibly cloned skb
Florian Westphal [Mon, 13 Mar 2017 15:24:28 +0000 (16:24 +0100)]
ipv6: avoid write to a possibly cloned skb

ip6_fragment, in case skb has a fraglist, checks if the
skb is cloned.  If it is, it will move to the 'slow path' and allocates
new skbs for each fragment.

However, right before entering the slowpath loop, it updates the
nexthdr value of the last ipv6 extension header to NEXTHDR_FRAGMENT,
to account for the fragment header that will be inserted in the new
ipv6-fragment skbs.

In case original skb is cloned this munges nexthdr value of another
skb.  Avoid this by doing the nexthdr update for each of the new fragment
skbs separately.

This was observed with tcpdump on a bridge device where netfilter ipv6
reassembly is active:  tcpdump shows malformed fragment headers as
the l4 header (icmpv6, tcp, etc). is decoded as a fragment header.

Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reported-by: Andreas Karis <akaris@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: wimax/i2400m: fix NULL-deref at probe
Johan Hovold [Mon, 13 Mar 2017 12:42:03 +0000 (13:42 +0100)]
net: wimax/i2400m: fix NULL-deref at probe

Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer or accessing memory beyond the endpoint array should a
malicious device lack the expected endpoints.

The endpoints are specifically dereferenced in the i2400m_bootrom_init
path during probe (e.g. in i2400mu_tx_bulk_out).

Fixes: f398e4240fce ("i2400m/USB: probe/disconnect, dev init/shutdown
and reset backends")
Cc: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoisdn/gigaset: fix NULL-deref at probe
Johan Hovold [Mon, 13 Mar 2017 12:39:01 +0000 (13:39 +0100)]
isdn/gigaset: fix NULL-deref at probe

Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer should a malicious device lack endpoints.

Fixes: cf7776dc05b8 ("[PATCH] isdn4linux: Siemens Gigaset drivers -
direct USB connection")
Cc: stable <stable@vger.kernel.org> # 2.6.17
Cc: Hansjoerg Lipp <hjlipp@web.de>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipv6: make ECMP route replacement less greedy
Sabrina Dubroca [Mon, 13 Mar 2017 12:28:09 +0000 (13:28 +0100)]
ipv6: make ECMP route replacement less greedy

Commit 27596472473a ("ipv6: fix ECMP route replacement") introduced a
loop that removes all siblings of an ECMP route that is being
replaced. However, this loop doesn't stop when it has replaced
siblings, and keeps removing other routes with a higher metric.
We also end up triggering the WARN_ON after the loop, because after
this nsiblings < 0.

Instead, stop the loop when we have taken care of all routes with the
same metric as the route being replaced.

  Reproducer:
  ===========
    #!/bin/sh

    ip netns add ns1
    ip netns add ns2
    ip -net ns1 link set lo up

    for x in 0 1 2 ; do
        ip link add veth$x netns ns2 type veth peer name eth$x netns ns1
        ip -net ns1 link set eth$x up
        ip -net ns2 link set veth$x up
    done

    ip -net ns1 -6 r a 2000::/64 nexthop via fe80::0 dev eth0 \
            nexthop via fe80::1 dev eth1 nexthop via fe80::2 dev eth2
    ip -net ns1 -6 r a 2000::/64 via fe80::42 dev eth0 metric 256
    ip -net ns1 -6 r a 2000::/64 via fe80::43 dev eth0 metric 2048

    echo "before replace, 3 routes"
    ip -net ns1 -6 r | grep -v '^fe80\|^ff00'
    echo

    ip -net ns1 -6 r c 2000::/64 nexthop via fe80::4 dev eth0 \
            nexthop via fe80::5 dev eth1 nexthop via fe80::6 dev eth2

    echo "after replace, only 2 routes, metric 2048 is gone"
    ip -net ns1 -6 r | grep -v '^fe80\|^ff00'

Fixes: 27596472473a ("ipv6: fix ECMP route replacement")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Reviewed-by: Xin Long <lucien.xin@gmail.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomm, gup: fix typo in gup_p4d_range()
Kirill A. Shutemov [Mon, 13 Mar 2017 05:22:13 +0000 (08:22 +0300)]
mm, gup: fix typo in gup_p4d_range()

gup_p4d_range() should call gup_pud_range(), not itself.

[ This was not noticed on x86: this is the HAVE_GENERIC_RCU_GUP code
  used by arm[64] and powerpc    - Linus ]

Fixes: c2febafc6773 ("mm: convert generic code to 5-level paging")
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reported-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reported-by: Anton Blanchard <anton@samba.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agobpf: improve read-only handling
Daniel Borkmann [Sat, 11 Mar 2017 15:55:49 +0000 (16:55 +0100)]
bpf: improve read-only handling

Improve bpf_{prog,jit_binary}_{un,}lock_ro() by throwing a
one-time warning in case of an error when the image couldn't
be set read-only, and also mark struct bpf_prog as locked when
bpf_prog_lock_ro() was called.

Reason for the latter is that bpf_prog_unlock_ro() is called from
various places including error paths, and we shouldn't mess with
page attributes when really not needed.

For bpf_jit_binary_unlock_ro() this is not needed as jited flag
implicitly indicates this, thus for archs with ARCH_HAS_SET_MEMORY
we're guaranteed to have a previously locked image. Overall, this
should also help us to identify any further potential issues with
set_memory_*() helpers.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoselftests/bpf: fix broken build
Alexei Starovoitov [Sat, 11 Mar 2017 06:05:55 +0000 (22:05 -0800)]
selftests/bpf: fix broken build

Recent merge of 'linux-kselftest-4.11-rc1' tree broke bpf test build.
None of the tests were building and test_verifier.c had tons of compiler errors.
Fix it and add #ifdef CAP_IS_SUPPORTED to support old versions of libcap.
Tested on centos 6.8 and 7

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agompls: Do not decrement alive counter for unregister events
David Ahern [Fri, 10 Mar 2017 22:11:39 +0000 (14:11 -0800)]
mpls: Do not decrement alive counter for unregister events

Multipath routes can be rendered usesless when a device in one of the
paths is deleted. For example:

$ ip -f mpls ro ls
100
nexthop as to 200 via inet 172.16.2.2  dev virt12
nexthop as to 300 via inet 172.16.3.2  dev br0
101
nexthop as to 201 via inet6 2000:2::2  dev virt12
nexthop as to 301 via inet6 2000:3::2  dev br0

$ ip li del br0

When br0 is deleted the other hop is not considered in
mpls_select_multipath because of the alive check -- rt_nhn_alive
is 0.

rt_nhn_alive is decremented once in mpls_ifdown when the device is taken
down (NETDEV_DOWN) and again when it is deleted (NETDEV_UNREGISTER). For
a 2 hop route, deleting one device drops the alive count to 0. Since
devices are taken down before unregistering, the decrement on
NETDEV_UNREGISTER is redundant.

Fixes: c89359a42e2a4 ("mpls: support for dead routes")
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoxen-netback: fix race condition on XenBus disconnect
Igor Druzhinin [Fri, 10 Mar 2017 21:36:22 +0000 (21:36 +0000)]
xen-netback: fix race condition on XenBus disconnect

In some cases during XenBus disconnect event handling and subsequent
queue resource release there may be some TX handlers active on
other processors. Use RCU in order to synchronize with them.

Signed-off-by: Igor Druzhinin <igor.druzhinin@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agompls: Send route delete notifications when router module is unloaded
David Ahern [Fri, 10 Mar 2017 17:46:15 +0000 (09:46 -0800)]
mpls: Send route delete notifications when router module is unloaded

When the mpls_router module is unloaded, mpls routes are deleted but
notifications are not sent to userspace leaving userspace caches
out of sync. Add the call to mpls_notify_route in mpls_net_exit as
routes are freed.

Fixes: 0189197f44160 ("mpls: Basic routing support")
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoact_connmark: avoid crashing on malformed nlattrs with null parms
Etienne Noss [Fri, 10 Mar 2017 15:55:32 +0000 (16:55 +0100)]
act_connmark: avoid crashing on malformed nlattrs with null parms

tcf_connmark_init does not check in its configuration if TCA_CONNMARK_PARMS
is set, resulting in a null pointer dereference when trying to access it.

[501099.043007] BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
[501099.043039] IP: [<ffffffffc10c60fb>] tcf_connmark_init+0x8b/0x180 [act_connmark]
...
[501099.044334] Call Trace:
[501099.044345]  [<ffffffffa47270e8>] ? tcf_action_init_1+0x198/0x1b0
[501099.044363]  [<ffffffffa47271b0>] ? tcf_action_init+0xb0/0x120
[501099.044380]  [<ffffffffa47250a4>] ? tcf_exts_validate+0xc4/0x110
[501099.044398]  [<ffffffffc0f5fa97>] ? u32_set_parms+0xa7/0x270 [cls_u32]
[501099.044417]  [<ffffffffc0f60bf0>] ? u32_change+0x680/0x87b [cls_u32]
[501099.044436]  [<ffffffffa4725d1d>] ? tc_ctl_tfilter+0x4dd/0x8a0
[501099.044454]  [<ffffffffa44a23a1>] ? security_capable+0x41/0x60
[501099.044471]  [<ffffffffa470ca01>] ? rtnetlink_rcv_msg+0xe1/0x220
[501099.044490]  [<ffffffffa470c920>] ? rtnl_newlink+0x870/0x870
[501099.044507]  [<ffffffffa472cc61>] ? netlink_rcv_skb+0xa1/0xc0
[501099.044524]  [<ffffffffa47073f4>] ? rtnetlink_rcv+0x24/0x30
[501099.044541]  [<ffffffffa472c634>] ? netlink_unicast+0x184/0x230
[501099.044558]  [<ffffffffa472c9d8>] ? netlink_sendmsg+0x2f8/0x3b0
[501099.044576]  [<ffffffffa46d8880>] ? sock_sendmsg+0x30/0x40
[501099.044592]  [<ffffffffa46d8e03>] ? SYSC_sendto+0xd3/0x150
[501099.044608]  [<ffffffffa425fda1>] ? __do_page_fault+0x2d1/0x510
[501099.044626]  [<ffffffffa47fbd7b>] ? system_call_fast_compare_end+0xc/0x9b

Fixes: 22a5dc0e5e3e ("net: sched: Introduce connmark action")
Signed-off-by: Étienne Noss <etienne.noss@wifirst.fr>
Signed-off-by: Victorien Molle <victorien.molle@wifirst.fr>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMake IP 'forwarding' doc more precise
Neil Jerram [Fri, 10 Mar 2017 12:24:57 +0000 (12:24 +0000)]
Make IP 'forwarding' doc more precise

It wasn't clear if the 'forwarding' setting needs to be enabled on the
interface that packets are received from, or on the interface that
packets are forwarded to, or both.

In fact (according to my code reading) the setting is relevant on the
interface that packets are received from, so this change updates the doc
to say that.

Signed-off-by: Neil Jerram <neil@tigera.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: handle select_queue when device is being removed
stephen hemminger [Thu, 9 Mar 2017 22:58:29 +0000 (14:58 -0800)]
netvsc: handle select_queue when device is being removed

Move the send indirection table from the inner device (netvsc)
to the network device context.

It is possible that netvsc_device is not present (remove in progress).
This solves potential use after free issues when packet is being
created during MTU change, shutdown, or queue count changes.

Fixes: d8e18ee0fa96 ("netvsc: enhance transmit select_queue")
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>