unionfs-2.6.39.y.git
10 years agoUSB: Nokia 305 should be treated as unusual dev
Victor A. Santos [Sun, 27 Apr 2014 02:20:14 +0000 (23:20 -0300)]
USB: Nokia 305 should be treated as unusual dev

commit f0ef5d41792a46a1085dead9dfb0bdb2c574638e upstream.

Signed-off-by: Victor A. Santos <victoraur.santos@gmail.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agousb: storage: shuttle_usbat: fix discs being detected twice
Daniele Forsi [Tue, 29 Apr 2014 09:44:03 +0000 (11:44 +0200)]
usb: storage: shuttle_usbat: fix discs being detected twice

commit df602c2d2358f02c6e49cffc5b49b9daa16db033 upstream.

Even if the USB-to-ATAPI converter supported multiple LUNs, this
driver would always detect the same physical device or media because
it doesn't use srb->device->lun in any way.
Tested with an Hewlett-Packard CD-Writer Plus 8200e.

Signed-off-by: Daniele Forsi <dforsi@gmail.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoUSB: OHCI: fix problem with global suspend on ATI controllers
Alan Stern [Thu, 1 May 2014 19:21:42 +0000 (15:21 -0400)]
USB: OHCI: fix problem with global suspend on ATI controllers

commit c1db30a2a79eb59997b13b8cabf2a50bea9f04e1 upstream.

Some OHCI controllers from ATI/AMD seem to have difficulty with
"global" USB suspend, that is, suspending an entire USB bus without
setting the suspend feature for each port connected to a device.  When
we try to resume the child devices, the controller gives timeout
errors on the unsuspended ports, requiring resets, and can even cause
ohci-hcd to hang; see

http://marc.info/?l=linux-usb&m=139514332820398&w=2

and the following messages.

This patch fixes the problem by adding a new quirk flag to ohci-hcd.
The flag causes the ohci_rh_suspend() routine to suspend each
unsuspended, enabled port before suspending the root hub.  This
effectively converts the "global" suspend to an ordinary root-hub
suspend.  There is no need to unsuspend these ports when the root hub
is resumed, because the child devices will be resumed anyway in the
course of a normal system resume ("global" suspend is never used for
runtime PM).

This patch should be applied to all stable kernels which include
commit 0aa2832dd0d9 (USB: use "global suspend" for system sleep on
USB-2 buses) or a backported version thereof.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Peter Münster <pmlists@free.fr>
Tested-by: Peter Münster <pmlists@free.fr>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agousb: gadget: at91-udc: fix irq and iomem resource retrieval
Jean-Jacques Hiblot [Wed, 12 Mar 2014 16:30:08 +0000 (17:30 +0100)]
usb: gadget: at91-udc: fix irq and iomem resource retrieval

commit 886c7c426d465732ec9d1b2bbdda5642fc2e7e05 upstream.

When using dt resources retrieval (interrupts and reg properties) there is
no predefined order for these resources in the platform dev resource
table. Also don't expect the number of resource to be always 2.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@traphandler.com>
Acked-by: Boris BREZILLON <b.brezillon@overkiz.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofsl-usb: do not test for PHY_CLK_VALID bit on controller version 1.6
Nikita Yushchenko [Mon, 28 Apr 2014 15:23:44 +0000 (19:23 +0400)]
fsl-usb: do not test for PHY_CLK_VALID bit on controller version 1.6

commit d183c81929beeba842b74422f754446ef2b8b49c upstream.

Per reference manuals of Freescale P1020 and P2020 SoCs, USB controller
present in these SoCs has bit 17 of USBx_CONTROL register marked as
Reserved - there is no PHY_CLK_VALID bit there.

Testing for this bit in ehci_fsl_setup_phy() behaves differently on two
P1020RDB boards available here - on one board test passes and fsl-usb
init succeeds, but on other board test fails, causing fsl-usb init to
fail.

This patch changes ehci_fsl_setup_phy() not to test PHY_CLK_VALID on
controller version 1.6 that (per manual) does not have this bit.

Signed-off-by: Nikita Yushchenko <nyushchenko@dev.rtsoft.ru>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiio:imu:mpu6050: Fixed segfault in Invensens MPU driver due to null dereference
Atilla Filiz [Fri, 11 Apr 2014 14:51:23 +0000 (16:51 +0200)]
iio:imu:mpu6050: Fixed segfault in Invensens MPU driver due to null dereference

commit b9b3a41893c3f1be67b5aacfa525969914bea0e9 upstream.

The driver segfaults when the kernel boots with device tree as the
platform data is then not present and the pointer is deferenced without
checking it is not null.  This patch introduces such a check avoiding the
crash.

Signed-off-by: Atilla Filiz <atilla.filiz@essensium.com>
Signed-off-by: Jonathan Cameron <jic23@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoclk: tegra: Fix wrong value written to PLLE_AUX
Tuomas Tynkkynen [Fri, 16 May 2014 13:50:20 +0000 (16:50 +0300)]
clk: tegra: Fix wrong value written to PLLE_AUX

commit d2c834abe2b39a2d5a6c38ef44de87c97cbb34b4 upstream.

The value written to PLLE_AUX was incorrect due to a wrong variable
being used. Without this fix SATA does not work.

Signed-off-by: Tuomas Tynkkynen <ttynkkynen@nvidia.com>
Tested-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Mike Turquette <mturquette@linaro.org>
[mturquette@linaro.org: improved changelog]
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/tegra: Remove gratuitous pad field
Thierry Reding [Wed, 9 Apr 2014 12:26:59 +0000 (14:26 +0200)]
drm/tegra: Remove gratuitous pad field

commit cbfbbabb89b37f6bad05f478d906a385149f288d upstream.

The version of the drm_tegra_submit structure that was merged all the
way back in 3.10 contains a pad field that was originally intended to
properly pad the following __u64 field. Unfortunately it seems like a
different field was dropped during review that caused this padding to
become unnecessary, but the pad field wasn't removed at that time.

One possible side-effect of this is that since the __u64 following the
pad is now no longer properly aligned, the compiler may (or may not)
introduce padding itself, which results in no predictable ABI.

Rectify this by removing the pad field so that all fields are again
naturally aligned. Technically this is breaking existing userspace ABI,
but given that there aren't any (released) userspace drivers that make
use of this yet, the fallout should be minimal.

Fixes: d43f81cbaf43 ("drm/tegra: Add gr2d device")
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: check buffer relocation offset
Leo Liu [Mon, 28 Apr 2014 13:40:22 +0000 (09:40 -0400)]
drm/radeon: check buffer relocation offset

commit 695daf1a8e731a4b5b89de89a61f32a4d7ad7094 upstream.

Signed-off-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: use pflip irq on R600+ v2
Christian König [Wed, 23 Apr 2014 18:46:06 +0000 (20:46 +0200)]
drm/radeon: use pflip irq on R600+ v2

commit f5d636d2a74b755879feec35e14a259de52ccc07 upstream.

Testing the update pending bit directly after issuing an
update is nonsense cause depending on the pixel clock the
CRTC needs a bit of time to execute the flip even when we
are in the VBLANK period.

This is just a non invasive patch to solve the problem at
hand, a more complete and cleaner solution should follow
in the next merge window.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=76564
v2: fix source IDs for CRTC2-6

Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon/uvd: use lower clocks on old UVD to boot v2
Christian König [Thu, 10 Apr 2014 14:11:36 +0000 (16:11 +0200)]
drm/radeon/uvd: use lower clocks on old UVD to boot v2

commit e45187620f9fc103edf68fa5ea78e73033e1668c upstream.

Some RV7xx generation hardware crashes after you
raise the UVD clocks for the first time. Try to
avoid this by using the lower clocks to boot these.

Workaround for: https://bugzilla.kernel.org/show_bug.cgi?id=71891

v2: lower clocks on IB test as well

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: fix count in cik_sdma_ring_test()
Alex Deucher [Tue, 22 Apr 2014 12:17:18 +0000 (08:17 -0400)]
drm/radeon: fix count in cik_sdma_ring_test()

commit 7e95cfb0b797678cd3493ca0322ef2675547a0bc upstream.

Should be 5 rather than 4.

Noticed-by: Mathias Fröhlich <Mathias.Froehlich@gmx.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon/pm: don't walk the crtc list before it has been initialized (v2)
Alex Deucher [Tue, 15 Apr 2014 16:44:33 +0000 (12:44 -0400)]
drm/radeon/pm: don't walk the crtc list before it has been initialized (v2)

commit 3ed9a335cfc64b2c83545f341cdddf2347b12b97 upstream.

Avoids a crash in certain cases when thermal irqs are generated
before the display structures have been initialized.

v2: fix the vblank and vrefresh helpers as well

bug:
https://bugzilla.kernel.org/show_bug.cgi?id=73931

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: fix ATPX detection on non-VGA GPUs
Alex Deucher [Tue, 15 Apr 2014 16:44:34 +0000 (12:44 -0400)]
drm/radeon: fix ATPX detection on non-VGA GPUs

commit e9a4099a59cc598a44006059dd775c25e422b772 upstream.

Some newer PX laptops have the pci device class
set to DISPLAY_OTHER rather than DISPLAY_VGA.  This
properly detects ATPX on those laptops.

Based on a patch from: Pali Rohár <pali.rohar@gmail.com>

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: airlied@gmail.com
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon/si: make sure mc ucode is loaded before checking the size
Alex Deucher [Wed, 16 Apr 2014 13:42:22 +0000 (09:42 -0400)]
drm/radeon/si: make sure mc ucode is loaded before checking the size

commit 8c79bae6a30f606b7a4e17c994bc5f72f8fdaf11 upstream.

Avoid a possible segfault.

Noticed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: add support for newer mc ucode on SI (v2)
Alex Deucher [Fri, 11 Apr 2014 15:21:49 +0000 (11:21 -0400)]
drm/radeon: add support for newer mc ucode on SI (v2)

commit 1ebe92802eaf0569784dce843bc28a78842d236c upstream.

May fix stability issues with some newer cards.

v2: print out mc firmware version used and size

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: disable mclk dpm on R7 260X
Alex Deucher [Fri, 11 Apr 2014 02:29:03 +0000 (22:29 -0400)]
drm/radeon: disable mclk dpm on R7 260X

commit 57700ad1f2f21d5d7ab7ee0e58d11b5954852434 upstream.

Setting higher mclks seems to cause stability issues
on some R7 260X boards.  Disable it for now for stability
until we find a proper fix.

bug:
https://bugs.freedesktop.org/show_bug.cgi?id=75992

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/i915: Break encoder->crtc link separately in intel_sanitize_crtc()
Egbert Eich [Fri, 25 Apr 2014 08:56:22 +0000 (10:56 +0200)]
drm/i915: Break encoder->crtc link separately in intel_sanitize_crtc()

commit 7f1950fbb989e8fc5463b307e062b4529d51c862 upstream.

Depending on the SDVO output_flags SDVO may have multiple connectors
linking to the same encoder (in intel_connector->encoder->base).
Only one of those connectors should be active (ie link to the encoder
thru drm_connector->encoder).
If intel_connector_break_all_links() is called from intel_sanitize_crtc()
we may break the crtc connection of an encoder thru an inactive connector
in which case intel_connector_break_all_links() will not be called again
for the active connector if this happens to come later in the list due to:
    if (connector->encoder->base.crtc != &crtc->base)
                                 continue;
in intel_sanitize_crtc().
This will however leave the drm_connector->encoder linkage for this
active connector in place. Subsequently this will cause multiple
warnings in intel_connector_check_state() to trigger and the driver
will eventually die in drm_encoder_crtc_ok() (because of crtc == NULL).

To avoid this remove intel_connector_break_all_links() and move its
code to its two calling functions: intel_sanitize_crtc() and
intel_sanitize_encoder().
This allows to implement the link breaking more flexibly matching
the surrounding code: ie. in intel_sanitize_crtc() we can break the
crtc link separatly after the links to the encoders have been
broken which avoids above problem.

This regression has been introduced in:

commit 24929352481f085c5f85d4d4cbc919ddf106d381
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Mon Jul 2 20:28:59 2012 +0200

    drm/i915: read out the modeset hw state at load and resume time

so goes back to the very beginning of the modeset rework.

v2: This patch takes care of the concernes voiced by Chris Wilson
and Daniel Vetter that only breaking links if the drm_connector
is linked to an encoder may miss some links.
v3: move all encoder handling to encoder loop as suggested by
Daniel Vetter.

Signed-off-by: Egbert Eich <eich@suse.de>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/i915: Don't check gmch state on inherited configs
Daniel Vetter [Sun, 13 Apr 2014 10:00:33 +0000 (12:00 +0200)]
drm/i915: Don't check gmch state on inherited configs

commit 9953599bc02dbc1d3330e6a0bfc6c50e9dffcac6 upstream.

... our current modeset code isn't good enough yet to handle this. The
scenario is:

1. BIOS sets up a cloned config with lvds+external screen on the same
pipe, e.g. pipe B.

2. We read out that state for pipe B and assign the gmch_pfit state to
it.

3. The initial modeset switches the lvds to pipe A but due to lack of
atomic modeset we don't recompute the config of pipe B.

-> both pipes now claim (in the sw pipe config structure) to use the
gmch_pfit, which just won't work.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74081
Tested-by: max <manikulin@gmail.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomd: avoid possible spinning md thread at shutdown.
NeilBrown [Mon, 5 May 2014 23:36:08 +0000 (09:36 +1000)]
md: avoid possible spinning md thread at shutdown.

commit 0f62fb220aa4ebabe8547d3a9ce4a16d3c045f21 upstream.

If an md array with externally managed metadata (e.g. DDF or IMSM)
is in use, then we should not set safemode==2 at shutdown because:

1/ this is ineffective: user-space need to be involved in any 'safemode' handling,
2/ The safemode management code doesn't cope with safemode==2 on external metadata
   and md_check_recover enters an infinite loop.

Even at shutdown, an infinite-looping process can be problematic, so this
could cause shutdown to hang.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohrtimer: Set expiry time before switch_hrtimer_base()
Viresh Kumar [Mon, 12 May 2014 08:12:29 +0000 (13:42 +0530)]
hrtimer: Set expiry time before switch_hrtimer_base()

commit 84ea7fe37908254c3bd90910921f6e1045c1747a upstream.

switch_hrtimer_base() calls hrtimer_check_target() which ensures that
we do not migrate a timer to a remote cpu if the timer expires before
the current programmed expiry time on that remote cpu.

But __hrtimer_start_range_ns() calls switch_hrtimer_base() before the
new expiry time is set. So the sanity check in hrtimer_check_target()
is operating on stale or even uninitialized data.

Update expiry time before calling switch_hrtimer_base().

[ tglx: Rewrote changelog once again ]

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Cc: linaro-kernel@lists.linaro.org
Cc: linaro-networking@linaro.org
Cc: fweisbec@gmail.com
Cc: arvind.chauhan@arm.com
Link: http://lkml.kernel.org/r/81999e148745fc51bbcd0615823fbab9b2e87e23.1399882253.git.viresh.kumar@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohrtimer: Prevent remote enqueue of leftmost timers
Leon Ma [Wed, 30 Apr 2014 08:43:10 +0000 (16:43 +0800)]
hrtimer: Prevent remote enqueue of leftmost timers

commit 012a45e3f4af68e86d85cce060c6c2fed56498b2 upstream.

If a cpu is idle and starts an hrtimer which is not pinned on that
same cpu, the nohz code might target the timer to a different cpu.

In the case that we switch the cpu base of the timer we already have a
sanity check in place, which determines whether the timer is earlier
than the current leftmost timer on the target cpu. In that case we
enqueue the timer on the current cpu because we cannot reprogram the
clock event device on the target.

If the timers base is already the target CPU we do not have this
sanity check in place so we enqueue the timer as the leftmost timer in
the target cpus rb tree, but we cannot reprogram the clock event
device on the target cpu. So the timer expires late and subsequently
prevents the reprogramming of the target cpu clock event device until
the previously programmed event fires or a timer with an earlier
expiry time gets enqueued on the target cpu itself.

Add the same target check as we have for the switch base case and
start the timer on the current cpu if it would become the leftmost
timer on the target.

[ tglx: Rewrote subject and changelog ]

Signed-off-by: Leon Ma <xindong.ma@intel.com>
Link: http://lkml.kernel.org/r/1398847391-5994-1-git-send-email-xindong.ma@intel.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohrtimer: Prevent all reprogramming if hang detected
Stuart Hayes [Tue, 29 Apr 2014 22:55:02 +0000 (17:55 -0500)]
hrtimer: Prevent all reprogramming if hang detected

commit 6c6c0d5a1c949d2e084706f9e5fb1fccc175b265 upstream.

If the last hrtimer interrupt detected a hang it sets hang_detected=1
and programs the clock event device with a delay to let the system
make progress.

If hang_detected == 1, we prevent reprogramming of the clock event
device in hrtimer_reprogram() but not in hrtimer_force_reprogram().

This can lead to the following situation:

hrtimer_interrupt()
   hang_detected = 1;
   program ce device to Xms from now (hang delay)

We have two timers pending:
   T1 expires 50ms from now
   T2 expires 5s from now

Now T1 gets canceled, which causes hrtimer_force_reprogram() to be
invoked, which in turn programs the clock event device to T2 (5
seconds from now).

Any hrtimer_start after that will not reprogram the hardware due to
hang_detected still being set. So we effectivly block all timers until
the T2 event fires and cleans up the hang situation.

Add a check for hang_detected to hrtimer_force_reprogram() which
prevents the reprogramming of the hang delay in the hardware
timer. The subsequent hrtimer_interrupt will resolve all outstanding
issues.

[ tglx: Rewrote subject and changelog and fixed up the comment in
   hrtimer_force_reprogram() ]

Signed-off-by: Stuart Hayes <stuart.w.hayes@gmail.com>
Link: http://lkml.kernel.org/r/53602DC6.2060101@gmail.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrivercore: deferral race condition fix
Grant Likely [Tue, 29 Apr 2014 11:05:22 +0000 (12:05 +0100)]
drivercore: deferral race condition fix

commit 58b116bce13612e5aa6fcd49ecbd4cf8bb59e835 upstream.

When the kernel is built with CONFIG_PREEMPT it is possible to reach a state
when all modules loaded but some driver still stuck in the deferred list
and there is a need for external event to kick the deferred queue to probe
these drivers.

The issue has been observed on embedded systems with CONFIG_PREEMPT enabled,
audio support built as modules and using nfsroot for root filesystem.

The following log fragment shows such sequence when all audio modules
were loaded but the sound card is not present since the machine driver has
failed to probe due to missing dependency during it's probe.
The board is am335x-evmsk (McASP<->tlv320aic3106 codec) with davinci-evm
machine driver:

...
[   12.615118] davinci-mcasp 4803c000.mcasp: davinci_mcasp_probe: ENTER
[   12.719969] davinci_evm sound.3: davinci_evm_probe: ENTER
[   12.725753] davinci_evm sound.3: davinci_evm_probe: snd_soc_register_card
[   12.753846] davinci-mcasp 4803c000.mcasp: davinci_mcasp_probe: snd_soc_register_component
[   12.922051] davinci-mcasp 4803c000.mcasp: davinci_mcasp_probe: snd_soc_register_component DONE
[   12.950839] davinci_evm sound.3: ASoC: platform (null) not registered
[   12.957898] davinci_evm sound.3: davinci_evm_probe: snd_soc_register_card DONE (-517)
[   13.099026] davinci-mcasp 4803c000.mcasp: Kicking the deferred list
[   13.177838] davinci-mcasp 4803c000.mcasp: really_probe: probe_count = 2
[   13.194130] davinci_evm sound.3: snd_soc_register_card failed (-517)
[   13.346755] davinci_mcasp_driver_init: LEAVE
[   13.377446] platform sound.3: Driver davinci_evm requests probe deferral
[   13.592527] platform sound.3: really_probe: probe_count = 0

In the log the machine driver enters it's probe at 12.719969 (this point it
has been removed from the deferred lists). McASP driver already executing
it's probing (since 12.615118).
The machine driver tries to construct the sound card (12.950839) but did
not found one of the components so it fails. After this McASP driver
registers all the ASoC components (the machine driver still in it's probe
function after it failed to construct the card) and the deferred work is
prepared at 13.099026 (note that this time the machine driver is not in the
lists so it is not going to be handled when the work is executing).
Lastly the machine driver exit from it's probe and the core places it to
the deferred list but there will be no other driver going to load and the
deferred queue is not going to be kicked again - till we have external event
like connecting USB stick, etc.

The proposed solution is to try the deferred queue once more when the last
driver is asking for deferring and we had drivers loaded while this last
driver was probing.

This way we can avoid drivers stuck in the deferred queue.

Signed-off-by: Grant Likely <grant.likely@linaro.org>
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Mark Brown <broonie@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohwmon: (emc1403) Support full range of known chip revision numbers
Josef Gajdusek [Mon, 12 May 2014 11:48:26 +0000 (13:48 +0200)]
hwmon: (emc1403) Support full range of known chip revision numbers

commit 3a18e1398fc2dc9c32bbdc50664da3a77959a8d1 upstream.

The datasheet for EMC1413/EMC1414, which is fully compatible to
EMC1403/1404 and uses the same chip identification, references revision
numbers 0x01, 0x03, and 0x04. Accept the full range of revision numbers
from 0x01 to 0x04 to make sure none are missed.

Signed-off-by: Josef Gajdusek <atx@atx.name>
[Guenter Roeck: Updated headline and description]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohwmon: (emc1403) fix inverted store_hyst()
Josef Gajdusek [Sun, 11 May 2014 12:40:44 +0000 (14:40 +0200)]
hwmon: (emc1403) fix inverted store_hyst()

commit 17c048fc4bd95efea208a1920f169547d8588f1f upstream.

Attempts to set the hysteresis value to a temperature below the target
limit fails with "write error: Numerical result out of range" due to
an inverted comparison.

Signed-off-by: Josef Gajdusek <atx@atx.name>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
[Guenter Roeck: Updated headline and description]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoaio: fix potential leak in aio_run_iocb().
Leon Yu [Thu, 1 May 2014 03:31:28 +0000 (03:31 +0000)]
aio: fix potential leak in aio_run_iocb().

commit 754320d6e166d3a12cb4810a452bde00afbd4e9a upstream.

iovec should be reclaimed whenever caller of rw_copy_check_uvector() returns,
but it doesn't hold when failure happens right after aio_setup_vectored_rw().

Fix that in a such way to avoid hairy goto.

Signed-off-by: Leon Yu <chianglungyu@gmail.com>
Signed-off-by: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agohwpoison, hugetlb: lock_page/unlock_page does not match for handling a free hugepage
Chen Yucong [Thu, 22 May 2014 18:54:15 +0000 (11:54 -0700)]
hwpoison, hugetlb: lock_page/unlock_page does not match for handling a free hugepage

commit b985194c8c0a130ed155b71662e39f7eaea4876f upstream.

For handling a free hugepage in memory failure, the race will happen if
another thread hwpoisoned this hugepage concurrently.  So we need to
check PageHWPoison instead of !PageHWPoison.

If hwpoison_filter(p) returns true or a race happens, then we need to
unlock_page(hpage).

Signed-off-by: Chen Yucong <slaoub@gmail.com>
Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Tested-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoarm64: fix pud_huge() for 2-level pagetables
Mark Salter [Thu, 15 May 2014 14:19:22 +0000 (15:19 +0100)]
arm64: fix pud_huge() for 2-level pagetables

commit 4797ec2dc83a43be35bad56037d1b53db9e2b5d5 upstream.

The following happens when trying to run a kvm guest on a kernel
configured for 64k pages. This doesn't happen with 4k pages:

  BUG: failure at include/linux/mm.h:297/put_page_testzero()!
  Kernel panic - not syncing: BUG!
  CPU: 2 PID: 4228 Comm: qemu-system-aar Tainted: GF            3.13.0-0.rc7.31.sa2.k32v1.aarch64.debug #1
  Call trace:
  [<fffffe0000096034>] dump_backtrace+0x0/0x16c
  [<fffffe00000961b4>] show_stack+0x14/0x1c
  [<fffffe000066e648>] dump_stack+0x84/0xb0
  [<fffffe0000668678>] panic+0xf4/0x220
  [<fffffe000018ec78>] free_reserved_area+0x0/0x110
  [<fffffe000018edd8>] free_pages+0x50/0x88
  [<fffffe00000a759c>] kvm_free_stage2_pgd+0x30/0x40
  [<fffffe00000a5354>] kvm_arch_destroy_vm+0x18/0x44
  [<fffffe00000a1854>] kvm_put_kvm+0xf0/0x184
  [<fffffe00000a1938>] kvm_vm_release+0x10/0x1c
  [<fffffe00001edc1c>] __fput+0xb0/0x288
  [<fffffe00001ede4c>] ____fput+0xc/0x14
  [<fffffe00000d5a2c>] task_work_run+0xa8/0x11c
  [<fffffe0000095c14>] do_notify_resume+0x54/0x58

In arch/arm/kvm/mmu.c:unmap_range(), we end up doing an extra put_page()
on the stage2 pgd which leads to the BUG in put_page_testzero(). This
happens because a pud_huge() test in unmap_range() returns true when it
should always be false with 2-level pages tables used by 64k pages.
This patch removes support for huge puds if 2-level pagetables are
being used.

Signed-off-by: Mark Salter <msalter@redhat.com>
[catalin.marinas@arm.com: removed #ifndef around PUD_SIZE check]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86, mm, hugetlb: Add missing TLB page invalidation for hugetlb_cow()
Anthony Iliopoulos [Wed, 14 May 2014 09:29:48 +0000 (11:29 +0200)]
x86, mm, hugetlb: Add missing TLB page invalidation for hugetlb_cow()

commit 9844f5462392b53824e8b86726e7c33b5ecbb676 upstream.

The invalidation is required in order to maintain proper semantics
under CoW conditions. In scenarios where a process clones several
threads, a thread operating on a core whose DTLB entry for a
particular hugepage has not been invalidated, will be reading from
the hugepage that belongs to the forked child process, even after
hugetlb_cow().

The thread will not see the updated page as long as the stale DTLB
entry remains cached, the thread attempts to write into the page,
the child process exits, or the thread gets migrated to a different
processor.

Signed-off-by: Anthony Iliopoulos <anthony.iliopoulos@huawei.com>
Link: http://lkml.kernel.org/r/20140514092948.GA17391@server-36.huawei.corp
Suggested-by: Shay Goikhman <shay.goikhman@huawei.com>
Acked-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm, thp: close race between mremap() and split_huge_page()
Kirill A. Shutemov [Fri, 9 May 2014 22:37:00 +0000 (15:37 -0700)]
mm, thp: close race between mremap() and split_huge_page()

commit dd18dbc2d42af75fffa60c77e0f02220bc329829 upstream.

It's critical for split_huge_page() (and migration) to catch and freeze
all PMDs on rmap walk.  It gets tricky if there's concurrent fork() or
mremap() since usually we copy/move page table entries on dup_mm() or
move_page_tables() without rmap lock taken.  To get it work we rely on
rmap walk order to not miss any entry.  We expect to see destination VMA
after source one to work correctly.

But after switching rmap implementation to interval tree it's not always
possible to preserve expected walk order.

It works fine for dup_mm() since new VMA has the same vma_start_pgoff()
/ vma_last_pgoff() and explicitly insert dst VMA after src one with
vma_interval_tree_insert_after().

But on move_vma() destination VMA can be merged into adjacent one and as
result shifted left in interval tree.  Fortunately, we can detect the
situation and prevent race with rmap walk by moving page table entries
under rmap lock.  See commit 38a76013ad80.

Problem is that we miss the lock when we move transhuge PMD.  Most
likely this bug caused the crash[1].

[1] http://thread.gmane.org/gmane.linux.kernel.mm/96473

Fixes: 108d6642ad81 ("mm anon rmap: remove anon_vma_moveto_tail")
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Dave Jones <davej@redhat.com>
Cc: David Miller <davem@davemloft.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: fix suspend vs. association race
Emmanuel Grumbach [Tue, 13 May 2014 09:54:09 +0000 (12:54 +0300)]
mac80211: fix suspend vs. association race

commit c52666aef9f2dff39276eb53f15d99e2e229870f upstream.

If the association is in progress while we suspend, the
stack will be in a messed up state. Clean it before we
suspend.

This patch completes Johannes's patch:

1a1cb744de160ee70086a77afff605bbc275d291
Author: Johannes Berg <johannes.berg@intel.com>

    mac80211: fix suspend vs. authentication race

Fixes: 12e7f517029d ("mac80211: cleanup generic suspend/resume procedures")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agocfg80211: free sme on connection failures
Eliad Peller [Wed, 30 Apr 2014 12:58:13 +0000 (15:58 +0300)]
cfg80211: free sme on connection failures

commit c1fbb258846dfc425507a093922d2d001e54c3ea upstream.

cfg80211 is notified about connection failures by
__cfg80211_connect_result() call. However, this
function currently does not free cfg80211 sme.

This results in hanging connection attempts in some cases

e.g. when mac80211 authentication attempt is denied,
we have this function call:
ieee80211_rx_mgmt_auth() -> cfg80211_rx_mlme_mgmt() ->
cfg80211_process_auth() -> cfg80211_sme_rx_auth() ->
__cfg80211_connect_result()

but cfg80211_sme_free() is never get called.

Fixes: ceca7b712 ("cfg80211: separate internal SME implementation")
Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipmi: Reset the KCS timeout when starting error recovery
Corey Minyard [Mon, 14 Apr 2014 14:46:52 +0000 (09:46 -0500)]
ipmi: Reset the KCS timeout when starting error recovery

commit eb6d78ec213e6938559b801421d64714dafcf4b2 upstream.

The OBF timer in KCS was not reset in one situation when error recovery
was started, resulting in an immediate timeout.

Reported-by: Bodo Stroesser <bstroesser@ts.fujitsu.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotimer: Prevent overflow in apply_slack
Jiri Bohac [Fri, 18 Apr 2014 15:23:11 +0000 (17:23 +0200)]
timer: Prevent overflow in apply_slack

commit 98a01e779f3c66b0b11cd7e64d531c0e41c95762 upstream.

On architectures with sizeof(int) < sizeof (long), the
computation of mask inside apply_slack() can be undefined if the
computed bit is > 32.

E.g. with: expires = 0xffffe6f5 and slack = 25, we get:

expires_limit = 0x20000000e
bit = 33
mask = (1 << 33) - 1  /* undefined */

On x86, mask becomes 1 and and the slack is not applied properly.
On s390, mask is -1, expires is set to 0 and the timer fires immediately.

Use 1UL << bit to solve that issue.

Suggested-by: Deborah Townsend <dstownse@us.ibm.com>
Signed-off-by: Jiri Bohac <jbohac@suse.cz>
Link: http://lkml.kernel.org/r/20140418152310.GA13654@midget.suse.cz
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agogpu: host1x: handle the correct # of syncpt regs
Stephen Warren [Fri, 4 Apr 2014 22:31:05 +0000 (16:31 -0600)]
gpu: host1x: handle the correct # of syncpt regs

commit 22bbd5d949dc7fdd72a4e78e767fa09d8e54b446 upstream.

BIT_WORD() truncates rather than rounds, so the loops in
syncpt_thresh_isr() and _host1x_intr_disable_all_syncpt_intrs() use <=
rather than < in an attempt to process the correct number of registers
when rounding of the conversion of count of bits to count of words is
necessary. However, when rounding isn't necessary because the value is
already a multiple of the divisor (as is the case for all values of
nb_pts the code actually sees), this causes one too many registers to
be processed.

Solve this by using and explicit DIV_ROUND_UP() call, rather than
BIT_WORD(), and comparing with < rather than <=.

Fixes: 7ede0b0bf3e2 ("gpu: host1x: Add syncpoint wait and interrupts")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-By: Terje Bergstrom <tbergstrom@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years ago8250_core: Fix unwanted TX chars write
Loic Poulain [Thu, 24 Apr 2014 09:38:56 +0000 (11:38 +0200)]
8250_core: Fix unwanted TX chars write

commit b08c9c317e3f7764a91d522cd031639ba42b98cc upstream.

On transmit-hold-register empty, serial8250_tx_chars
should be called only if we don't use DMA.
DMA has its own tx cycle.

Signed-off-by: Loic Poulain <loic.poulain@intel.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoserial: 8250: Fix thread unsafe __dma_tx_complete function
Loic Poulain [Thu, 24 Apr 2014 09:34:48 +0000 (11:34 +0200)]
serial: 8250: Fix thread unsafe __dma_tx_complete function

commit f8fd1b0350d3a4581125f5eda6528f5a2c5f9183 upstream.

__dma_tx_complete is not protected against concurrent
call of serial8250_tx_dma. it can lead to circular tail
index corruption or parallel call of serial_tx_dma on the
same data portion.

This patch fixes this issue by holding the port lock.

Signed-off-by: Loic Poulain <loic.poulain@intel.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm: make fixup_user_fault() check the vma access rights too
Linus Torvalds [Tue, 22 Apr 2014 20:49:40 +0000 (13:49 -0700)]
mm: make fixup_user_fault() check the vma access rights too

commit 1b17844b29ae042576bea588164f2f1e9590a8bc upstream.

fixup_user_fault() is used by the futex code when the direct user access
fails, and the futex code wants it to either map in the page in a usable
form or return an error.  It relied on handle_mm_fault() to map the
page, and correctly checked the error return from that, but while that
does map the page, it doesn't actually guarantee that the page will be
mapped with sufficient permissions to be then accessed.

So do the appropriate tests of the vma access rights by hand.

[ Side note: arguably handle_mm_fault() could just do that itself, but
  we have traditionally done it in the caller, because some callers -
  notably get_user_pages() - have been able to access pages even when
  they are mapped with PROT_NONE.  Maybe we should re-visit that design
  decision, but in the meantime this is the minimal patch. ]

Found by Dave Jones running his trinity tool.

Reported-by: Dave Jones <davej@redhat.com>
Acked-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agocoredump: fix va_list corruption
Eric Dumazet [Sat, 19 Apr 2014 17:15:07 +0000 (10:15 -0700)]
coredump: fix va_list corruption

commit 404ca80eb5c2727d78cd517d12108b040c522e12 upstream.

A va_list needs to be copied in case it needs to be used twice.

Thanks to Hugh for debugging this issue, leading to various panics.

Tested:

  lpq84:~# echo "|/foobar12345 %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h" >/proc/sys/kernel/core_pattern

'produce_core' is simply : main() { *(int *)0 = 1;}

  lpq84:~# ./produce_core
  Segmentation fault (core dumped)
  lpq84:~# dmesg | tail -1
  [  614.352947] Core dump to |/foobar12345 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 lpq84 (null) pipe failed

Notice the last argument was replaced by a NULL (we were lucky enough to
not crash, but do not try this on your production machine !)

After fix :

  lpq83:~# echo "|/foobar12345 %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h %h" >/proc/sys/kernel/core_pattern
  lpq83:~# ./produce_core
  Segmentation fault
  lpq83:~# dmesg | tail -1
  [  740.800441] Core dump to |/foobar12345 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 lpq83 pipe failed

Fixes: 5fe9d8ca21cc ("coredump: cn_vprintf() has no reason to call vsnprintf() twice")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Diagnosed-by: Hugh Dickins <hughd@google.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agopata_at91: fix ata_host_activate() failure handling
Bartlomiej Zolnierkiewicz [Mon, 31 Mar 2014 17:51:14 +0000 (19:51 +0200)]
pata_at91: fix ata_host_activate() failure handling

commit 27aa64b9d1bd0d23fd692c91763a48309b694311 upstream.

Add missing clk_put() call to ata_host_activate() failure path.

Sergei says,

  "Hm, I have once fixed that (see that *if* (!ret)) but looks like a
   later commit 477c87e90853d136b188c50c0e4a93d01cad872e (ARM:
   at91/pata: use gpio_is_valid to check the gpio) broke it again. :-(
   Would be good if the changelog did mention that..."

Cc: Andrew Victor <linux@maxim.org.za>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoclocksource: Exynos_mct: Register clock event after request_irq()
Krzysztof Kozlowski [Wed, 16 Apr 2014 14:36:45 +0000 (14:36 +0000)]
clocksource: Exynos_mct: Register clock event after request_irq()

commit 8db6e5104b77de5d0b7002b95069da0992a34be9 upstream.

After hotplugging CPU1 the first call of interrupt handler for CPU1
oneshot timer was called on CPU0 because it fired before setting IRQ
affinity. Affected are SoCs where Multi Core Timer interrupts are
shared (SPI), e.g. Exynos 4210.

During setup of the MCT timers the clock event device should be
registered after setting the affinity for interrupt. This will prevent
starting the timer too early.

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>,
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
Cc: Kukjin Kim <kgene.kim@samsung.com>
Cc: linux-arm-kernel@lists.infradead.org,
Link: http://lkml.kernel.org/r/20140416143316.299247848@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoclocksource: Exynos_mct: Use irq_force_affinity() in cpu bringup
Thomas Gleixner [Wed, 16 Apr 2014 14:36:45 +0000 (14:36 +0000)]
clocksource: Exynos_mct: Use irq_force_affinity() in cpu bringup

commit 30ccf03b4a6a2102a2219058bdc6d779dc637dd7 upstream.

The starting cpu is not yet in the online mask so irq_set_affinity()
fails which results in per cpu timers for this cpu ending up on some
other online cpu, ususally cpu 0.

Use irq_force_affinity() which disables the online mask check and
makes things work.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>,
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
Cc: Kukjin Kim <kgene.kim@samsung.com>
Cc: linux-arm-kernel@lists.infradead.org,
Link: http://lkml.kernel.org/r/20140416143316.106665251@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agogenirq: Allow forcing cpu affinity of interrupts
Thomas Gleixner [Wed, 16 Apr 2014 14:36:44 +0000 (14:36 +0000)]
genirq: Allow forcing cpu affinity of interrupts

commit 01f8fa4f01d8362358eb90e412bd7ae18a3ec1ad upstream.

The current implementation of irq_set_affinity() refuses rightfully to
route an interrupt to an offline cpu.

But there is a special case, where this is actually desired. Some of
the ARM SoCs have per cpu timers which require setting the affinity
during cpu startup where the cpu is not yet in the online mask.

If we can't do that, then the local timer interrupt for the about to
become online cpu is routed to some random online cpu.

The developers of the affected machines tried to work around that
issue, but that results in a massive mess in that timer code.

We have a yet unused argument in the set_affinity callbacks of the irq
chips, which I added back then for a similar reason. It was never
required so it got not used. But I'm happy that I never removed it.

That allows us to implement a sane handling of the above scenario. So
the affected SoC drivers can add the required force handling to their
interrupt chip, switch the timer code to irq_force_affinity() and
things just work.

This does not affect any existing user of irq_set_affinity().

Tagged for stable to allow a simple fix of the affected SoC clock
event drivers.

Reported-and-tested-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>,
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
Cc: Kukjin Kim <kgene.kim@samsung.com>
Cc: linux-arm-kernel@lists.infradead.org,
Link: http://lkml.kernel.org/r/20140416143315.717251504@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoirqchip: Gic: Support forced affinity setting
Thomas Gleixner [Wed, 16 Apr 2014 14:36:44 +0000 (14:36 +0000)]
irqchip: Gic: Support forced affinity setting

commit ffde1de64012c406dfdda8690918248b472f24e4 upstream.

To support the affinity setting of per cpu timers in the early startup
of a not yet online cpu, implement the force logic, which disables the
cpu online check.

Tagged for stable to allow a simple fix of the affected SoC clock
event drivers.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>,
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
Cc: Kukjin Kim <kgene.kim@samsung.com>
Cc: linux-arm-kernel@lists.infradead.org,
Link: http://lkml.kernel.org/r/20140416143315.916984416@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoftrace/module: Hardcode ftrace_module_init() call into load_module()
Steven Rostedt (Red Hat) [Thu, 24 Apr 2014 14:40:12 +0000 (10:40 -0400)]
ftrace/module: Hardcode ftrace_module_init() call into load_module()

commit a949ae560a511fe4e3adf48fa44fefded93e5c2b upstream.

A race exists between module loading and enabling of function tracer.

CPU 1 CPU 2
----- -----
  load_module()
   module->state = MODULE_STATE_COMING

register_ftrace_function()
 mutex_lock(&ftrace_lock);
 ftrace_startup()
  update_ftrace_function();
   ftrace_arch_code_modify_prepare()
    set_all_module_text_rw();
   <enables-ftrace>
    ftrace_arch_code_modify_post_process()
     set_all_module_text_ro();

[ here all module text is set to RO,
  including the module that is
  loading!! ]

   blocking_notifier_call_chain(MODULE_STATE_COMING);
    ftrace_init_module()

     [ tries to modify code, but it's RO, and fails!
       ftrace_bug() is called]

When this race happens, ftrace_bug() will produces a nasty warning and
all of the function tracing features will be disabled until reboot.

The simple solution is to treate module load the same way the core
kernel is treated at boot. To hardcode the ftrace function modification
of converting calls to mcount into nops. This is done in init/main.c
there's no reason it could not be done in load_module(). This gives
a better control of the changes and doesn't tie the state of the
module to its notifiers as much. Ftrace is special, it needs to be
treated as such.

The reason this would work, is that the ftrace_module_init() would be
called while the module is in MODULE_STATE_UNFORMED, which is ignored
by the set_all_module_text_ro() call.

Link: http://lkml.kernel.org/r/1395637826-3312-1-git-send-email-indou.takao@jp.fujitsu.com
Reported-by: Takao Indoh <indou.takao@jp.fujitsu.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomips: dts: Fix missing device_type="memory" property in memory nodes
Leif Lindholm [Thu, 17 Apr 2014 17:42:00 +0000 (18:42 +0100)]
mips: dts: Fix missing device_type="memory" property in memory nodes

commit dfc44f8030653b345fc6fb337558c3a07536823f upstream.

A few platforms lack a 'device_type = "memory"' for their memory
nodes, relying on an old ppc quirk in order to discover its memory.
Add the missing data so that all parsing code can find memory nodes
correctly.

Signed-off-by: Leif Lindholm <leif.lindholm@linaro.org>
Cc: linux-mips@linux-mips.org
Cc: devicetree@vger.kernel.org
Cc: Mark Rutland <mark.rutland@arm.com>
Acked-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Grant Likely <grant.likely@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofutex: Prevent attaching to kernel threads
Thomas Gleixner [Mon, 12 May 2014 20:45:35 +0000 (20:45 +0000)]
futex: Prevent attaching to kernel threads

commit f0d71b3dcb8332f7971b5f2363632573e6d9486a upstream.

We happily allow userspace to declare a random kernel thread to be the
owner of a user space PI futex.

Found while analysing the fallout of Dave Jones syscall fuzzer.

We also should validate the thread group for private futexes and find
some fast way to validate whether the "alleged" owner has RW access on
the file which backs the SHM, but that's a separate issue.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Dave Jones <davej@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Darren Hart <darren@dvhart.com>
Cc: Davidlohr Bueso <davidlohr@hp.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Clark Williams <williams@redhat.com>
Cc: Paul McKenney <paulmck@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Carlos ODonell <carlos@redhat.com>
Cc: Jakub Jelinek <jakub@redhat.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: http://lkml.kernel.org/r/20140512201701.194824402@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofutex: Add another early deadlock detection check
Thomas Gleixner [Mon, 12 May 2014 20:45:34 +0000 (20:45 +0000)]
futex: Add another early deadlock detection check

commit 866293ee54227584ffcb4a42f69c1f365974ba7f upstream.

Dave Jones trinity syscall fuzzer exposed an issue in the deadlock
detection code of rtmutex:
  http://lkml.kernel.org/r/20140429151655.GA14277@redhat.com

That underlying issue has been fixed with a patch to the rtmutex code,
but the futex code must not call into rtmutex in that case because
    - it can detect that issue early
    - it avoids a different and more complex fixup for backing out

If the user space variable got manipulated to 0x80000000 which means
no lock holder, but the waiters bit set and an active pi_state in the
kernel is found we can figure out the recursive locking issue by
looking at the pi_state owner. If that is the current task, then we
can safely return -EDEADLK.

The check should have been added in commit 59fa62451 (futex: Handle
futex_pi OWNER_DIED take over correctly) already, but I did not see
the above issue caused by user space manipulation back then.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Dave Jones <davej@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Darren Hart <darren@dvhart.com>
Cc: Davidlohr Bueso <davidlohr@hp.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Clark Williams <williams@redhat.com>
Cc: Paul McKenney <paulmck@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Carlos ODonell <carlos@redhat.com>
Cc: Jakub Jelinek <jakub@redhat.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: http://lkml.kernel.org/r/20140512201701.097349971@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoACPI / video: Fix initial level validity test
Aaron Lu [Wed, 6 Nov 2013 01:03:15 +0000 (09:03 +0800)]
ACPI / video: Fix initial level validity test

commit 9efa5e50598c5568b0678bb411b239a0b6e9a328 upstream.

When testing if the firmware's initial value is valid, we should use
the corrected level value instead of the raw value returned from
firmware.

Signed-off-by: Aaron Lu <aaron.lu@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoclk: tegra: use pll_ref as the pll_e parent
Peter De Schrijver [Mon, 25 Nov 2013 12:44:13 +0000 (14:44 +0200)]
clk: tegra: use pll_ref as the pll_e parent

commit 8e9cc80aa348938078c3c1a7ab55efb3c40990e3 upstream.

Use pll_ref instead of pll_re_vco as the pll_e parent on Tegra114. Also
add a 12Mhz pll_ref table entry for pll_e for Tegra114. This prevents
the system from crashing at bootup because of an unsupported pll_re_vco
rate.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agousb: qcserial: add a number of Dell devices
Bjørn Mork [Sun, 27 Apr 2014 14:47:42 +0000 (16:47 +0200)]
usb: qcserial: add a number of Dell devices

commit 4d7c0136a54f62501f8a34c4d08a5e0258d3d3ca upstream.

Dan writes:

"The Dell drivers use the same configuration for PIDs:

81A2: Dell Wireless 5806 Gobi(TM) 4G LTE Mobile Broadband Card
81A3: Dell Wireless 5570 HSPA+ (42Mbps) Mobile Broadband Card
81A4: Dell Wireless 5570e HSPA+ (42Mbps) Mobile Broadband Card
81A8: Dell Wireless 5808 Gobi(TM) 4G LTE Mobile Broadband Card
81A9: Dell Wireless 5808e Gobi(TM) 4G LTE Mobile Broadband Card

These devices are all clearly Sierra devices, but are also definitely
Gobi-based.  The A8 might be the MC7700/7710 and A9 is likely a MC7750.

>From DellGobi5kSetup.exe from the Dell drivers:

usbif0: serial/firmware loader?
usbif2: nmea
usbif3: modem/ppp
usbif8: net/QMI"

Cc: <stable@vger.kernel.org>
Reported-by: AceLan Kao <acelan.kao@canonical.com>
Reported-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonetfilter: Fix potential use after free in ip6_route_me_harder()
Sergey Popovich [Thu, 8 May 2014 13:22:35 +0000 (16:22 +0300)]
netfilter: Fix potential use after free in ip6_route_me_harder()

commit a8951d5814e1373807a94f79f7ccec7041325470 upstream.

Dst is released one line before we access it again with dst->error.

Fixes: 58e35d147128 netfilter: ipv6: propagate routing errors from
ip6_route_me_harder()

Signed-off-by: Sergey Popovich <popovich_sergei@mail.ru>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: fix on-channel remain-on-channel
Johannes Berg [Wed, 14 May 2014 13:34:41 +0000 (15:34 +0200)]
mac80211: fix on-channel remain-on-channel

commit b4b177a5556a686909e643f1e9b6434c10de079f upstream.

Jouni reported that if a remain-on-channel was active on the
same channel as the current operating channel, then the ROC
would start, but any frames transmitted using mgmt-tx on the
same channel would get delayed until after the ROC.

The reason for this is that the ROC starts, but doesn't have
any handling for "remain on the same channel", so it stops
the interface queues. The later mgmt-tx then puts the frame
on the interface queues (since it's on the current operating
channel) and thus they get delayed until after the ROC.

To fix this, add some logic to handle remaining on the same
channel specially and not stop the queues etc. in this case.
This not only fixes the bug but also improves behaviour in
this case as data frames etc. can continue to flow.

Cc: stable@vger.kernel.org
Reported-by: Jouni Malinen <j@w1.fi>
Tested-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm, oom: prefer thread group leaders for display purposes
David Rientjes [Thu, 23 Jan 2014 23:53:34 +0000 (15:53 -0800)]
mm, oom: prefer thread group leaders for display purposes

commit d49ad9355420c743c736bfd1dee9eaa5b1a7722a upstream.

When two threads have the same badness score, it's preferable to kill
the thread group leader so that the actual process name is printed to
the kernel log rather than the thread group name which may be shared
amongst several processes.

This was the behavior when select_bad_process() used to do
for_each_process(), but it now iterates threads instead and leads to
ambiguity.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Greg Thelen <gthelen@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agooom_kill: add rcu_read_lock() into find_lock_task_mm()
Oleg Nesterov [Tue, 21 Jan 2014 23:50:01 +0000 (15:50 -0800)]
oom_kill: add rcu_read_lock() into find_lock_task_mm()

commit 4d4048be8a93769350efa31d2482a038b7de73d0 upstream.

find_lock_task_mm() expects it is called under rcu or tasklist lock, but
it seems that at least oom_unkillable_task()->task_in_mem_cgroup() and
mem_cgroup_out_of_memory()->oom_badness() can call it lockless.

Perhaps we could fix the callers, but this patch simply adds rcu lock
into find_lock_task_mm().  This also allows to simplify a bit one of its
callers, oom_kill_process().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Sergey Dyasly <dserrg@gmail.com>
Cc: Sameer Nanda <snanda@chromium.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: "Ma, Xindong" <xindong.ma@intel.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: "Tu, Xiaobing" <xiaobing.tu@intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agooom_kill: has_intersects_mems_allowed() needs rcu_read_lock()
Oleg Nesterov [Tue, 21 Jan 2014 23:50:00 +0000 (15:50 -0800)]
oom_kill: has_intersects_mems_allowed() needs rcu_read_lock()

commit ad96244179fbd55b40c00f10f399bc04739b8e1f upstream.

At least out_of_memory() calls has_intersects_mems_allowed() without
even rcu_read_lock(), this is obviously buggy.

Add the necessary rcu_read_lock().  This means that we can not simply
return from the loop, we need "bool ret" and "break".

While at it, swap the names of task_struct's (the argument and the
local).  This cleans up the code a little bit and avoids the unnecessary
initialization.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Sergey Dyasly <dserrg@gmail.com>
Tested-by: Sergey Dyasly <dserrg@gmail.com>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: "Ma, Xindong" <xindong.ma@intel.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: "Tu, Xiaobing" <xiaobing.tu@intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agooom_kill: change oom_kill.c to use for_each_thread()
Oleg Nesterov [Tue, 21 Jan 2014 23:49:58 +0000 (15:49 -0800)]
oom_kill: change oom_kill.c to use for_each_thread()

commit 1da4db0cd5c8a31d4468ec906b413e75e604b465 upstream.

Change oom_kill.c to use for_each_thread() rather than the racy
while_each_thread() which can loop forever if we race with exit.

Note also that most users were buggy even if while_each_thread() was
fine, the task can exit even _before_ rcu_read_lock().

Fortunately the new for_each_thread() only requires the stable
task_struct, so this change fixes both problems.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Sergey Dyasly <dserrg@gmail.com>
Tested-by: Sergey Dyasly <dserrg@gmail.com>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: "Ma, Xindong" <xindong.ma@intel.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: "Tu, Xiaobing" <xiaobing.tu@intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agointroduce for_each_thread() to replace the buggy while_each_thread()
Oleg Nesterov [Tue, 21 Jan 2014 23:49:56 +0000 (15:49 -0800)]
introduce for_each_thread() to replace the buggy while_each_thread()

commit 0c740d0afc3bff0a097ad03a1c8df92757516f5c upstream.

while_each_thread() and next_thread() should die, almost every lockless
usage is wrong.

1. Unless g == current, the lockless while_each_thread() is not safe.

   while_each_thread(g, t) can loop forever if g exits, next_thread()
   can't reach the unhashed thread in this case. Note that this can
   happen even if g is the group leader, it can exec.

2. Even if while_each_thread() itself was correct, people often use
   it wrongly.

   It was never safe to just take rcu_read_lock() and loop unless
   you verify that pid_alive(g) == T, even the first next_thread()
   can point to the already freed/reused memory.

This patch adds signal_struct->thread_head and task->thread_node to
create the normal rcu-safe list with the stable head.  The new
for_each_thread(g, t) helper is always safe under rcu_read_lock() as
long as this task_struct can't go away.

Note: of course it is ugly to have both task_struct->thread_node and the
old task_struct->thread_group, we will kill it later, after we change
the users of while_each_thread() to use for_each_thread().

Perhaps we can kill it even before we convert all users, we can
reimplement next_thread(t) using the new thread_head/thread_node.  But
we can't do this right now because this will lead to subtle behavioural
changes.  For example, do/while_each_thread() always sees at least one
task, while for_each_thread() can do nothing if the whole thread group
has died.  Or thread_group_empty(), currently its semantics is not clear
unless thread_group_leader(p) and we need to audit the callers before we
can change it.

So this patch adds the new interface which has to coexist with the old
one for some time, hopefully the next changes will be more or less
straightforward and the old one will go away soon.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Sergey Dyasly <dserrg@gmail.com>
Tested-by: Sergey Dyasly <dserrg@gmail.com>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: "Ma, Xindong" <xindong.ma@intel.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: "Tu, Xiaobing" <xiaobing.tu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoLinux 3.12.21 v3.12.21
Jiri Slaby [Thu, 29 May 2014 13:47:24 +0000 (15:47 +0200)]
Linux 3.12.21

10 years agonet-gro: reset skb->truesize in napi_reuse_skb()
Eric Dumazet [Thu, 3 Apr 2014 16:28:10 +0000 (09:28 -0700)]
net-gro: reset skb->truesize in napi_reuse_skb()

[ Upstream commit e33d0ba8047b049c9262fdb1fcafb93cb52ceceb ]

Recycling skb always had been very tough...

This time it appears GRO layer can accumulate skb->truesize
adjustments made by drivers when they attach a fragment to skb.

skb_gro_receive() can only subtract from skb->truesize the used part
of a fragment.

I spotted this problem seeing TcpExtPruneCalled and
TcpExtTCPRcvCollapsed that were unexpected with a recent kernel, where
TCP receive window should be sized properly to accept traffic coming
from a driver not overshooting skb->truesize.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv4: initialise the itag variable in __mkroute_input
Li RongQing [Thu, 22 May 2014 08:36:55 +0000 (16:36 +0800)]
ipv4: initialise the itag variable in __mkroute_input

[ Upstream commit fbdc0ad095c0a299e9abf5d8ac8f58374951149a ]

the value of itag is a random value from stack, and may not be initiated by
fib_validate_source, which called fib_combine_itag if CONFIG_IP_ROUTE_CLASSID
is not set

This will make the cached dst uncertainty

Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoip_tunnel: Initialize the fallback device properly
Steffen Klassert [Mon, 19 May 2014 09:36:56 +0000 (11:36 +0200)]
ip_tunnel: Initialize the fallback device properly

[ Upstream commit 78ff4be45a4c51d8fb21ad92e4fabb467c6c3eeb ]

We need to initialize the fallback device to have a correct mtu
set on this device. Otherwise the mtu is set to null and the device
is unusable.

Fixes: fd58156e456d ("IPIP: Use ip-tunneling code.")
Cc: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agobatman-adv: increase orig refcount when storing ref in gw_node
Antonio Quartulli [Thu, 1 May 2014 23:35:13 +0000 (01:35 +0200)]
batman-adv: increase orig refcount when storing ref in gw_node

[ Upstream commit 377fe0f968b30a1a714fab53a908061914f30e26 ]

A pointer to the orig_node representing a bat-gateway is
stored in the gw_node->orig_node member, but the refcount
for such orig_node is never increased.
This leads to memory faults when gw_node->orig_node is accessed
and the originator has already been freed.

Fix this by increasing the refcount on gw_node creation
and decreasing it on gw_node free.

Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agortnetlink: wait for unregistering devices in rtnl_link_unregister()
Cong Wang [Mon, 12 May 2014 22:11:20 +0000 (15:11 -0700)]
rtnetlink: wait for unregistering devices in rtnl_link_unregister()

[ Upstream commit 200b916f3575bdf11609cb447661b8d5957b0bbf ]

From: Cong Wang <cwang@twopensource.com>

commit 50624c934db18ab90 (net: Delay default_device_exit_batch until no
devices are unregistering) introduced rtnl_lock_unregistering() for
default_device_exit_batch(). Same race could happen we when rmmod a driver
which calls rtnl_link_unregister() as we call dev->destructor without rtnl
lock.

For long term, I think we should clean up the mess of netdev_run_todo()
and net namespce exit code.

Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoip6_tunnel: fix potential NULL pointer dereference
Susant Sahani [Fri, 9 May 2014 18:41:32 +0000 (00:11 +0530)]
ip6_tunnel: fix potential NULL pointer dereference

[ Upstream commit c8965932a2e3b70197ec02c6741c29460279e2a8 ]

The function ip6_tnl_validate assumes that the rtnl
attribute IFLA_IPTUN_PROTO always be filled . If this
attribute is not filled by  the userspace application
kernel get crashed with NULL pointer dereference. This
patch fixes the potential kernel crash when
IFLA_IPTUN_PROTO is missing .

Signed-off-by: Susant Sahani <susant@redhat.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosfc: fix calling of free_irq with already free vector
Nikolay Aleksandrov [Fri, 9 May 2014 09:11:39 +0000 (11:11 +0200)]
sfc: fix calling of free_irq with already free vector

[ Upstream commit 1c3639005f48492e5f2d965779efd814e80f8b15 ]

If the sfc driver is in legacy interrupt mode (either explicitly by
using interrupt_mode module param or by falling back to it) it will
hit a warning at kernel/irq/manage.c because it will try to free an irq
which wasn't allocated by it in the first place because the MSI(X) irqs are
zero and it'll try to free them unconditionally. So fix it by checking if
we're in legacy mode and freeing the appropriate irqs.

CC: Zenghui Shi <zshi@redhat.com>
CC: Ben Hutchings <ben@decadent.org.uk>
CC: <linux-net-drivers@solarflare.com>
CC: Shradha Shah <sshah@solarflare.com>
CC: David S. Miller <davem@davemloft.net>
Fixes: 1899c111a535 ("sfc: Fix IRQ cleanup in case of a probe failure")
Reported-by: Zenghui Shi <zshi@redhat.com>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Shradha Shah <sshah@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomacvlan: Don't propagate IFF_ALLMULTI changes on down interfaces.
Peter Christensen [Thu, 8 May 2014 09:15:37 +0000 (11:15 +0200)]
macvlan: Don't propagate IFF_ALLMULTI changes on down interfaces.

[ Upstream commit bbeb0eadcf9fe74fb2b9b1a6fea82cd538b1e556 ]

Clearing the IFF_ALLMULTI flag on a down interface could cause an allmulti
overflow on the underlying interface.

Attempting the set IFF_ALLMULTI on the underlying interface would cause an
error and the log message:

"allmulti touches root, set allmulti failed."

Signed-off-by: Peter Christensen <pch@ordbogen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: cdc_mbim: handle unaccelerated VLAN tagged frames
Bjørn Mork [Fri, 9 May 2014 12:45:00 +0000 (14:45 +0200)]
net: cdc_mbim: handle unaccelerated VLAN tagged frames

[ Upstream commit 6b5eeb7f874b689403e52a646e485d0191ab9507 ]

This driver maps 802.1q VLANs to MBIM sessions. The mapping is based on
a bogus assumption that all tagged frames will use the acceleration API
because we enable NETIF_F_HW_VLAN_CTAG_TX. This fails for e.g. frames
tagged in userspace using packet sockets. Such frames will erroneously
be considered as untagged and silently dropped based on not being IP.

Fix by falling back to looking into the ethernet header for a tag if no
accelerated tag was found.

Fixes: a82c7ce5bc5b ("net: cdc_ncm: map MBIM IPS SessionID to VLAN ID")
Cc: Greg Suarez <gsuarez@smithmicro.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv4: fib_semantics: increment fib_info_cnt after fib_info allocation
Sergey Popovich [Tue, 6 May 2014 15:23:08 +0000 (18:23 +0300)]
ipv4: fib_semantics: increment fib_info_cnt after fib_info allocation

[ Upstream commit aeefa1ecfc799b0ea2c4979617f14cecd5cccbfd ]

Increment fib_info_cnt in fib_create_info() right after successfuly
alllocating fib_info structure, overwise fib_metrics allocation failure
leads to fib_info_cnt incorrectly decremented in free_fib_info(), called
on error path from fib_create_info().

Signed-off-by: Sergey Popovich <popovich_sergei@mail.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: ipv6: send pkttoobig immediately if orig frag size > mtu
Florian Westphal [Sun, 4 May 2014 22:03:34 +0000 (00:03 +0200)]
net: ipv6: send pkttoobig immediately if orig frag size > mtu

[ Upstream commit 418a31561d594a2b636c1e2fa94ecd9e1245abb1 ]

If conntrack defragments incoming ipv6 frags it stores largest original
frag size in ip6cb and sets ->local_df.

We must thus first test the largest original frag size vs. mtu, and not
vice versa.

Without this patch PKTTOOBIG is still generated in ip6_fragment() later
in the stack, but

1) IPSTATS_MIB_INTOOBIGERRORS won't increment
2) packet did (needlessly) traverse netfilter postrouting hook.

Fixes: fe6cc55f3a9 ("net: ip, ipv6: handle gso skbs in forwarding path")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: ipv4: ip_forward: fix inverted local_df test
Florian Westphal [Sun, 4 May 2014 21:24:31 +0000 (23:24 +0200)]
net: ipv4: ip_forward: fix inverted local_df test

[ Upstream commit ca6c5d4ad216d5942ae544bbf02503041bd802aa ]

local_df means 'ignore DF bit if set', so if its set we're
allowed to perform ip fragmentation.

This wasn't noticed earlier because the output path also drops such skbs
(and emits needed icmp error) and because netfilter ip defrag did not
set local_df until couple of days ago.

Only difference is that DF-packets-larger-than MTU now discarded
earlier (f.e. we avoid pointless netfilter postrouting trip).

While at it, drop the repeated test ip_exceeds_mtu, checking it once
is enough...

Fixes: fe6cc55f3a9 ("net: ip, ipv6: handle gso skbs in forwarding path")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoip_tunnel: Set network header properly for IP_ECN_decapsulate()
Ying Cai [Sun, 4 May 2014 22:20:04 +0000 (15:20 -0700)]
ip_tunnel: Set network header properly for IP_ECN_decapsulate()

[ Upstream commit e96f2e7c430014eff52c93cabef1ad4f42ed0db1 ]

In ip_tunnel_rcv(), set skb->network_header to inner IP header
before IP_ECN_decapsulate().

Without the fix, IP_ECN_decapsulate() takes outer IP header as
inner IP header, possibly causing error messages or packet drops.

Note that this skb_reset_network_header() call was in this spot when
the original feature for checking consistency of ECN bits through
tunnels was added in eccc1bb8d4b4 ("tunnel: drop packet if ECN present
with not-ECT"). It was only removed from this spot in 3d7b46cd20e3
("ip_tunnel: push generic protocol handling to ip_tunnel module.").

Fixes: 3d7b46cd20e3 ("ip_tunnel: push generic protocol handling to ip_tunnel module.")
Reported-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Ying Cai <ycai@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agovsock: Make transport the proto owner
Andy King [Thu, 1 May 2014 22:20:43 +0000 (15:20 -0700)]
vsock: Make transport the proto owner

[ Upstream commit 2c4a336e0a3e203fab6aa8d8f7bb70a0ad968a6b ]

Right now the core vsock module is the owner of the proto family. This
means there's nothing preventing the transport module from unloading if
there are open sockets, which results in a panic. Fix that by allowing
the transport to be the owner, which will refcount it properly.

Includes version bump to 1.0.1.0-k

Passes checkpatch this time, I swear...

Acked-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotcp_cubic: fix the range of delayed_ack
Liu Yu [Wed, 30 Apr 2014 09:34:09 +0000 (17:34 +0800)]
tcp_cubic: fix the range of delayed_ack

[ Upstream commit 0cda345d1b2201dd15591b163e3c92bad5191745 ]

commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
divide error) try to prevent divide error, but there is still a little
chance that delayed_ack can reach zero. In case the param cnt get
negative value, then ratio+cnt would overflow and may happen to be zero.
As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.

In some old kernels, such as 2.6.32, there is a bug that would
pass negative param, which then ultimately leads to this divide error.

commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
with skb MSS) fixed the negative param issue. However,
it's safe that we fix the range of delayed_ack as well,
to make sure we do not hit a divide by zero.

CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Liu Yu <allanyuliu@tencent.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoRevert "macvlan : fix checksums error when we are in bridge mode"
Vlad Yasevich [Tue, 29 Apr 2014 14:09:51 +0000 (10:09 -0400)]
Revert "macvlan : fix checksums error when we are in bridge mode"

[ Upstream commit f114890cdf84d753f6b41cd0cc44ba51d16313da ]

This reverts commit 12a2856b604476c27d85a5f9a57ae1661fc46019.
The commit above doesn't appear to be necessary any more as the
checksums appear to be correctly computed/validated.

Additionally the above commit breaks kvm configurations where
one VM is using a device that support checksum offload (virtio) and
the other VM does not.
In this case, packets leaving virtio device will have CHECKSUM_PARTIAL
set.  The packets is forwarded to a macvtap that has offload features
turned off.  Since we use CHECKSUM_UNNECESSARY, the host does does not
update the checksum and thus a bad checksum is passed up to
the guest.

CC: Daniel Lezcano <daniel.lezcano@free.fr>
CC: Patrick McHardy <kaber@trash.net>
CC: Andrian Nord <nightnord@gmail.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Jason Wang <jasowang@redhat.com>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomactap: Fix checksum errors for non-gso packets in bridge mode
Vlad Yasevich [Tue, 29 Apr 2014 14:09:50 +0000 (10:09 -0400)]
mactap: Fix checksum errors for non-gso packets in bridge mode

[ Upstream commit cbdb04279ccaefcc702c8757757eea8ed76e50cf ]

The following is a problematic configuration:

 VM1: virtio-net device connected to macvtap0@eth0
 VM2: e1000 device connect to macvtap1@eth0

The problem is is that virtio-net supports checksum offloading
and thus sends the packets to the host with CHECKSUM_PARTIAL set.
On the other hand, e1000 does not support any acceleration.

For small TCP packets (and this includes the 3-way handshake),
e1000 ends up receiving packets that only have a partial checksum
set.  This causes TCP to fail checksum validation and to drop
packets.  As a result tcp connections can not be established.

Commit 3e4f8b787370978733ca6cae452720a4f0c296b8
macvtap: Perform GSO on forwarding path.
fixes this issue for large packets wthat will end up undergoing GSO.
This commit adds a check for the non-GSO case and attempts to
compute the checksum for partially checksummed packets in the
non-GSO case.

CC: Daniel Lezcano <daniel.lezcano@free.fr>
CC: Patrick McHardy <kaber@trash.net>
CC: Andrian Nord <nightnord@gmail.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Jason Wang <jasowang@redhat.com>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoslip: fix spinlock variant
Oliver Hartkopp [Sat, 26 Apr 2014 19:18:32 +0000 (21:18 +0200)]
slip: fix spinlock variant

[ Upstream commit ddcde142bed44490e338ed1124cb149976d355bb ]

With commit cc9fa74e2a ("slip/slcan: added locking in wakeup function") a
formerly missing locking was added to slip.c and slcan.c by Andre Naujoks.

Alexander Stein contributed the fix 367525c8c2 ("can: slcan: Fix spinlock
variant") as the kernel lock debugging advised to use spin_lock_bh() instead
of just using spin_lock().

This fix has to be applied to the same code section in slip.c for the same
reason too.

Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add a number of Dell devices
Bjørn Mork [Fri, 25 Apr 2014 17:00:34 +0000 (19:00 +0200)]
net: qmi_wwan: add a number of Dell devices

[ Upstream commit 6f10c5d1b1aeddb63d33070abb8bc5a177beeb1f ]

Dan writes:

"The Dell drivers use the same configuration for PIDs:

81A2: Dell Wireless 5806 Gobi(TM) 4G LTE Mobile Broadband Card
81A3: Dell Wireless 5570 HSPA+ (42Mbps) Mobile Broadband Card
81A4: Dell Wireless 5570e HSPA+ (42Mbps) Mobile Broadband Card
81A8: Dell Wireless 5808 Gobi(TM) 4G LTE Mobile Broadband Card
81A9: Dell Wireless 5808e Gobi(TM) 4G LTE Mobile Broadband Card

These devices are all clearly Sierra devices, but are also definitely
Gobi-based.  The A8 might be the MC7700/7710 and A9 is likely a MC7750.

>From DellGobi5kSetup.exe from the Dell drivers:

usbif0: serial/firmware loader?
usbif2: nmea
usbif3: modem/ppp
usbif8: net/QMI"

Reported-by: AceLan Kao <acelan.kao@canonical.com>
Reported-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add a number of CMOTech devices
Bjørn Mork [Fri, 25 Apr 2014 17:00:33 +0000 (19:00 +0200)]
net: qmi_wwan: add a number of CMOTech devices

[ Upstream commit 41be7d90993b1502d445bfc59e58348c258ce66a ]

A number of older CMOTech modems are based on Qualcomm
chips and exporting a QMI/wwan function.

Reported-by: Lars Melin <larsm17@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add Alcatel L800MA
Bjørn Mork [Fri, 25 Apr 2014 17:00:32 +0000 (19:00 +0200)]
net: qmi_wwan: add Alcatel L800MA

[ Upstream commit 75573660c47a0db7cc931dcf154945610e02130a ]

Device interface layout:
0: ff/ff/ff - serial
1: ff/00/00 - serial AT+PPP
2: ff/ff/ff - QMI/wwan
3: 08/06/50 - storage

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add Olivetti Olicard 500
Bjørn Mork [Fri, 25 Apr 2014 17:00:31 +0000 (19:00 +0200)]
net: qmi_wwan: add Olivetti Olicard 500

[ Upstream commit efc0b25c3add97717ece57bf5319792ca98f348e ]

Device interface layout:
0: ff/ff/ff - serial
1: ff/ff/ff - serial AT+PPP
2: 08/06/50 - storage
3: ff/ff/ff - serial
4: ff/ff/ff - QMI/wwan

Reported-by: Julio Araujo <julio.araujo@wllctel.com.br>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add Sierra Wireless MC7305/MC7355
Bjørn Mork [Fri, 25 Apr 2014 17:00:30 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless MC7305/MC7355

[ Upstream commit 9214224e43e4264b02686ea8b455f310935607b5 ]

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add Sierra Wireless MC73xx
Bjørn Mork [Fri, 25 Apr 2014 17:00:29 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless MC73xx

[ Upstream commit 1c138607a7be64074d7fba68d0d533ec38f9d17b ]

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: qmi_wwan: add Sierra Wireless EM7355
Bjørn Mork [Fri, 25 Apr 2014 17:00:28 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless EM7355

[ Upstream commit b85f5deaf052340021d025e120a9858f084a1d79 ]

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosctp: reset flowi4_oif parameter on route lookup
Xufeng Zhang [Fri, 25 Apr 2014 08:55:41 +0000 (16:55 +0800)]
sctp: reset flowi4_oif parameter on route lookup

[ Upstream commit 85350871317a5adb35519d9dc6fc9e80809d42ad ]

commit 813b3b5db83 (ipv4: Use caller's on-stack flowi as-is
in output route lookups.) introduces another regression which
is very similar to the problem of commit e6b45241c (ipv4: reset
flowi parameters on route connect) wants to fix:
Before we call ip_route_output_key() in sctp_v4_get_dst() to
get a dst that matches a bind address as the source address,
we have already called this function previously and the flowi
parameters have been initialized including flowi4_oif, so when
we call this function again, the process in __ip_route_output_key()
will be different because of the setting of flowi4_oif, and we'll
get a networking device which corresponds to the inputted flowi4_oif
as the output device, this is wrong because we'll never hit this
place if the previously returned source address of dst match one
of the bound addresses.

To reproduce this problem, a vlan setting is enough:
  # ifconfig eth0 up
  # route del default
  # vconfig add eth0 2
  # vconfig add eth0 3
  # ifconfig eth0.2 10.0.1.14 netmask 255.255.255.0
  # route add default gw 10.0.1.254 dev eth0.2
  # ifconfig eth0.3 10.0.0.14 netmask 255.255.255.0
  # ip rule add from 10.0.0.14 table 4
  # ip route add table 4 default via 10.0.0.254 src 10.0.0.14 dev eth0.3
  # sctp_darn -H 10.0.0.14 -P 36422 -h 10.1.4.134 -p 36422 -s -I
You'll detect that all the flow are routed to eth0.2(10.0.1.254).

Signed-off-by: Xufeng Zhang <xufeng.zhang@windriver.com>
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agobridge: Handle IFLA_ADDRESS correctly when creating bridge device
Toshiaki Makita [Fri, 25 Apr 2014 08:01:18 +0000 (17:01 +0900)]
bridge: Handle IFLA_ADDRESS correctly when creating bridge device

[ Upstream commit 30313a3d5794472c3548d7288e306a5492030370 ]

When bridge device is created with IFLA_ADDRESS, we are not calling
br_stp_change_bridge_id(), which leads to incorrect local fdb
management and bridge id calculation, and prevents us from receiving
frames on the bridge device.

Reported-by: Tom Gundersen <teg@jklm.no>
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv6: fib: fix fib dump restart
Kumar Sundararajan [Thu, 24 Apr 2014 13:48:53 +0000 (09:48 -0400)]
ipv6: fib: fix fib dump restart

[ Upstream commit 1c2658545816088477e91860c3a645053719cb54 ]

When the ipv6 fib changes during a table dump, the walk is
restarted and the number of nodes dumped are skipped. But the existing
code doesn't advance to the next node after a node is skipped. This can
cause the dump to loop or produce lots of duplicates when the fib
is modified during the dump.

This change advances the walk to the next node if the current node is
skipped after a restart.

Signed-off-by: Kumar Sundararajan <kumar@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agortnetlink: Only supply IFLA_VF_PORTS information when RTEXT_FILTER_VF is set
David Gibson [Thu, 24 Apr 2014 00:22:36 +0000 (10:22 +1000)]
rtnetlink: Only supply IFLA_VF_PORTS information when RTEXT_FILTER_VF is set

[ Upstream commit c53864fd60227de025cb79e05493b13f69843971 ]

Since 115c9b81928360d769a76c632bae62d15206a94a (rtnetlink: Fix problem with
buffer allocation), RTM_NEWLINK messages only contain the IFLA_VFINFO_LIST
attribute if they were solicited by a GETLINK message containing an
IFLA_EXT_MASK attribute with the RTEXT_FILTER_VF flag.

That was done because some user programs broke when they received more data
than expected - because IFLA_VFINFO_LIST contains information for each VF
it can become large if there are many VFs.

However, the IFLA_VF_PORTS attribute, supplied for devices which implement
ndo_get_vf_port (currently the 'enic' driver only), has the same problem.
It supplies per-VF information and can therefore become large, but it is
not currently conditional on the IFLA_EXT_MASK value.

Worse, it interacts badly with the existing EXT_MASK handling.  When
IFLA_EXT_MASK is not supplied, the buffer for netlink replies is fixed at
NLMSG_GOODSIZE.  If the information for IFLA_VF_PORTS exceeds this, then
rtnl_fill_ifinfo() returns -EMSGSIZE on the first message in a packet.
netlink_dump() will misinterpret this as having finished the listing and
omit data for this interface and all subsequent ones.  That can cause
getifaddrs(3) to enter an infinite loop.

This patch addresses the problem by only supplying IFLA_VF_PORTS when
IFLA_EXT_MASK is supplied with the RTEXT_FILTER_VF flag set.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agortnetlink: Warn when interface's information won't fit in our packet
David Gibson [Thu, 24 Apr 2014 00:22:35 +0000 (10:22 +1000)]
rtnetlink: Warn when interface's information won't fit in our packet

[ Upstream commit 973462bbde79bb827824c73b59027a0aed5c9ca6 ]

Without IFLA_EXT_MASK specified, the information reported for a single
interface in response to RTM_GETLINK is expected to fit within a netlink
packet of NLMSG_GOODSIZE.

If it doesn't, however, things will go badly wrong,  When listing all
interfaces, netlink_dump() will incorrectly treat -EMSGSIZE on the first
message in a packet as the end of the listing and omit information for
that interface and all subsequent ones.  This can cause getifaddrs(3) to
enter an infinite loop.

This patch won't fix the problem, but it will WARN_ON() making it easier to
track down what's going wrong.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: Fix ns_capable check in sock_diag_put_filterinfo
Andrew Lutomirski [Thu, 17 Apr 2014 04:41:34 +0000 (21:41 -0700)]
net: Fix ns_capable check in sock_diag_put_filterinfo

[ Upstream commit 78541c1dc60b65ecfce5a6a096fc260219d6784e ]

The caller needs capabilities on the namespace being queried, not on
their own namespace.  This is a security bug, although it likely has
only a minor impact.

Cc: stable@vger.kernel.org
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: sctp: cache auth_enable per endpoint
Vlad Yasevich [Thu, 17 Apr 2014 15:26:50 +0000 (17:26 +0200)]
net: sctp: cache auth_enable per endpoint

[ Upstream commit b14878ccb7fac0242db82720b784ab62c467c0dc ]

Currently, it is possible to create an SCTP socket, then switch
auth_enable via sysctl setting to 1 and crash the system on connect:

Oops[#1]:
CPU: 0 PID: 0 Comm: swapper Not tainted 3.14.1-mipsgit-20140415 #1
task: ffffffff8056ce80 ti: ffffffff8055c000 task.ti: ffffffff8055c000
[...]
Call Trace:
[<ffffffff8043c4e8>] sctp_auth_asoc_set_default_hmac+0x68/0x80
[<ffffffff8042b300>] sctp_process_init+0x5e0/0x8a4
[<ffffffff8042188c>] sctp_sf_do_5_1B_init+0x234/0x34c
[<ffffffff804228c8>] sctp_do_sm+0xb4/0x1e8
[<ffffffff80425a08>] sctp_endpoint_bh_rcv+0x1c4/0x214
[<ffffffff8043af68>] sctp_rcv+0x588/0x630
[<ffffffff8043e8e8>] sctp6_rcv+0x10/0x24
[<ffffffff803acb50>] ip6_input+0x2c0/0x440
[<ffffffff8030fc00>] __netif_receive_skb_core+0x4a8/0x564
[<ffffffff80310650>] process_backlog+0xb4/0x18c
[<ffffffff80313cbc>] net_rx_action+0x12c/0x210
[<ffffffff80034254>] __do_softirq+0x17c/0x2ac
[<ffffffff800345e0>] irq_exit+0x54/0xb0
[<ffffffff800075a4>] ret_from_irq+0x0/0x4
[<ffffffff800090ec>] rm7k_wait_irqoff+0x24/0x48
[<ffffffff8005e388>] cpu_startup_entry+0xc0/0x148
[<ffffffff805a88b0>] start_kernel+0x37c/0x398
Code: dd0900b8  000330f8  0126302d <dcc6000050c0fff1  0047182a  a48306a0
03e00008  00000000
---[ end trace b530b0551467f2fd ]---
Kernel panic - not syncing: Fatal exception in interrupt

What happens while auth_enable=0 in that case is, that
ep->auth_hmacs is initialized to NULL in sctp_auth_init_hmacs()
when endpoint is being created.

After that point, if an admin switches over to auth_enable=1,
the machine can crash due to NULL pointer dereference during
reception of an INIT chunk. When we enter sctp_process_init()
via sctp_sf_do_5_1B_init() in order to respond to an INIT chunk,
the INIT verification succeeds and while we walk and process
all INIT params via sctp_process_param() we find that
net->sctp.auth_enable is set, therefore do not fall through,
but invoke sctp_auth_asoc_set_default_hmac() instead, and thus,
dereference what we have set to NULL during endpoint
initialization phase.

The fix is to make auth_enable immutable by caching its value
during endpoint initialization, so that its original value is
being carried along until destruction. The bug seems to originate
from the very first days.

Fix in joint work with Daniel Borkmann.

Reported-by: Joshua Kinard <kumba@gentoo.org>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Tested-by: Joshua Kinard <kumba@gentoo.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotg3: update rx_jumbo_pending ring param only when jumbo frames are enabled
Ivan Vecera [Thu, 17 Apr 2014 12:51:08 +0000 (14:51 +0200)]
tg3: update rx_jumbo_pending ring param only when jumbo frames are enabled

The patch fixes a problem with dropped jumbo frames after usage of
'ethtool -G ... rx'.

Scenario:
1. ip link set eth0 up
2. ethtool -G eth0 rx N # <- This zeroes rx-jumbo
3. ip link set mtu 9000 dev eth0

The ethtool command set rx_jumbo_pending to zero so any received jumbo
packets are dropped and you need to use 'ethtool -G eth0 rx-jumbo N'
to workaround the issue.
The patch changes the logic so rx_jumbo_pending value is changed only if
jumbo frames are enabled (MTU > 1500).

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Acked-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agovlan: Fix lockdep warning when vlan dev handle notification
dingtianhong [Thu, 17 Apr 2014 10:40:36 +0000 (18:40 +0800)]
vlan: Fix lockdep warning when vlan dev handle notification

[ Upstream commit dc8eaaa006350d24030502a4521542e74b5cb39f ]

When I open the LOCKDEP config and run these steps:

modprobe 8021q
vconfig add eth2 20
vconfig add eth2.20 30
ifconfig eth2 xx.xx.xx.xx

then the Call Trace happened:

[32524.386288] =============================================
[32524.386293] [ INFO: possible recursive locking detected ]
[32524.386298] 3.14.0-rc2-0.7-default+ #35 Tainted: G           O
[32524.386302] ---------------------------------------------
[32524.386306] ifconfig/3103 is trying to acquire lock:
[32524.386310]  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff814275f4>] dev_mc_sync+0x64/0xb0
[32524.386326]
[32524.386326] but task is already holding lock:
[32524.386330]  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff8141af83>] dev_set_rx_mode+0x23/0x40
[32524.386341]
[32524.386341] other info that might help us debug this:
[32524.386345]  Possible unsafe locking scenario:
[32524.386345]
[32524.386350]        CPU0
[32524.386352]        ----
[32524.386354]   lock(&vlan_netdev_addr_lock_key/1);
[32524.386359]   lock(&vlan_netdev_addr_lock_key/1);
[32524.386364]
[32524.386364]  *** DEADLOCK ***
[32524.386364]
[32524.386368]  May be due to missing lock nesting notation
[32524.386368]
[32524.386373] 2 locks held by ifconfig/3103:
[32524.386376]  #0:  (rtnl_mutex){+.+.+.}, at: [<ffffffff81431d42>] rtnl_lock+0x12/0x20
[32524.386387]  #1:  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff8141af83>] dev_set_rx_mode+0x23/0x40
[32524.386398]
[32524.386398] stack backtrace:
[32524.386403] CPU: 1 PID: 3103 Comm: ifconfig Tainted: G           O 3.14.0-rc2-0.7-default+ #35
[32524.386409] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[32524.386414]  ffffffff81ffae40 ffff8800d9625ae8 ffffffff814f68a2 ffff8800d9625bc8
[32524.386421]  ffffffff810a35fb ffff8800d8a8d9d0 00000000d9625b28 ffff8800d8a8e5d0
[32524.386428]  000003cc00000000 0000000000000002 ffff8800d8a8e5f8 0000000000000000
[32524.386435] Call Trace:
[32524.386441]  [<ffffffff814f68a2>] dump_stack+0x6a/0x78
[32524.386448]  [<ffffffff810a35fb>] __lock_acquire+0x7ab/0x1940
[32524.386454]  [<ffffffff810a323a>] ? __lock_acquire+0x3ea/0x1940
[32524.386459]  [<ffffffff810a4874>] lock_acquire+0xe4/0x110
[32524.386464]  [<ffffffff814275f4>] ? dev_mc_sync+0x64/0xb0
[32524.386471]  [<ffffffff814fc07a>] _raw_spin_lock_nested+0x2a/0x40
[32524.386476]  [<ffffffff814275f4>] ? dev_mc_sync+0x64/0xb0
[32524.386481]  [<ffffffff814275f4>] dev_mc_sync+0x64/0xb0
[32524.386489]  [<ffffffffa0500cab>] vlan_dev_set_rx_mode+0x2b/0x50 [8021q]
[32524.386495]  [<ffffffff8141addf>] __dev_set_rx_mode+0x5f/0xb0
[32524.386500]  [<ffffffff8141af8b>] dev_set_rx_mode+0x2b/0x40
[32524.386506]  [<ffffffff8141b3cf>] __dev_open+0xef/0x150
[32524.386511]  [<ffffffff8141b177>] __dev_change_flags+0xa7/0x190
[32524.386516]  [<ffffffff8141b292>] dev_change_flags+0x32/0x80
[32524.386524]  [<ffffffff8149ca56>] devinet_ioctl+0x7d6/0x830
[32524.386532]  [<ffffffff81437b0b>] ? dev_ioctl+0x34b/0x660
[32524.386540]  [<ffffffff814a05b0>] inet_ioctl+0x80/0xa0
[32524.386550]  [<ffffffff8140199d>] sock_do_ioctl+0x2d/0x60
[32524.386558]  [<ffffffff81401a52>] sock_ioctl+0x82/0x2a0
[32524.386568]  [<ffffffff811a7123>] do_vfs_ioctl+0x93/0x590
[32524.386578]  [<ffffffff811b2705>] ? rcu_read_lock_held+0x45/0x50
[32524.386586]  [<ffffffff811b39e5>] ? __fget_light+0x105/0x110
[32524.386594]  [<ffffffff811a76b1>] SyS_ioctl+0x91/0xb0
[32524.386604]  [<ffffffff815057e2>] system_call_fastpath+0x16/0x1b

========================================================================

The reason is that all of the addr_lock_key for vlan dev have the same class,
so if we change the status for vlan dev, the vlan dev and its real dev will
hold the same class of addr_lock_key together, so the warning happened.

we should distinguish the lock depth for vlan dev and its real dev.

v1->v2: Convert the vlan_netdev_addr_lock_key to an array of eight elements, which
could support to add 8 vlan id on a same vlan dev, I think it is enough for current
scene, because a netdev's name is limited to IFNAMSIZ which could not hold 8 vlan id,
and the vlan dev would not meet the same class key with its real dev.

The new function vlan_dev_get_lockdep_subkey() will return the subkey and make the vlan
dev could get a suitable class key.

v2->v3: According David's suggestion, I use the subclass to distinguish the lock key for vlan dev
and its real dev, but it make no sense, because the difference for subclass in the
lock_class_key doesn't mean that the difference class for lock_key, so I use lock_depth
to distinguish the different depth for every vlan dev, the same depth of the vlan dev
could have the same lock_class_key, I import the MAX_LOCK_DEPTH from the include/linux/sched.h,
I think it is enough here, the lockdep should never exceed that value.

v3->v4: Add a huge array of locking keys will waste static kernel memory and is not a appropriate method,
we could use _nested() variants to fix the problem, calculate the depth for every vlan dev,
and use the depth as the subclass for addr_lock_key.

Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoip6_gre: don't allow to remove the fb_tunnel_dev
Nicolas Dichtel [Mon, 14 Apr 2014 15:11:38 +0000 (17:11 +0200)]
ip6_gre: don't allow to remove the fb_tunnel_dev

[ Upstream commit 54d63f787b652755e66eb4dd8892ee6d3f5197fc ]

It's possible to remove the FB tunnel with the command 'ip link del ip6gre0' but
this is unsafe, the module always supposes that this device exists. For example,
ip6gre_tunnel_lookup() may use it unconditionally.

Let's add a rtnl handler for dellink, which will never remove the FB tunnel (we
let ip6gre_destroy_tunnels() do the job).

Introduced by commit c12b395a4664 ("gre: Support GRE over IPv6").

CC: Dmitry Kozlov <xeb@mail.ru>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofilter: prevent nla extensions to peek beyond the end of the message
Mathias Krause [Sun, 13 Apr 2014 16:23:33 +0000 (18:23 +0200)]
filter: prevent nla extensions to peek beyond the end of the message

[ Upstream commit 05ab8f2647e4221cbdb3856dd7d32bd5407316b3 ]

The BPF_S_ANC_NLATTR and BPF_S_ANC_NLATTR_NEST extensions fail to check
for a minimal message length before testing the supplied offset to be
within the bounds of the message. This allows the subtraction of the nla
header to underflow and therefore -- as the data type is unsigned --
allowing far to big offset and length values for the search of the
netlink attribute.

The remainder calculation for the BPF_S_ANC_NLATTR_NEST extension is
also wrong. It has the minuend and subtrahend mixed up, therefore
calculates a huge length value, allowing to overrun the end of the
message while looking for the netlink attribute.

The following three BPF snippets will trigger the bugs when attached to
a UNIX datagram socket and parsing a message with length 1, 2 or 3.

 ,-[ PoC for missing size check in BPF_S_ANC_NLATTR ]--
 | ld #0x87654321
 | ldx #42
 | ld #nla
 | ret a
 `---

 ,-[ PoC for the same bug in BPF_S_ANC_NLATTR_NEST ]--
 | ld #0x87654321
 | ldx #42
 | ld #nlan
 | ret a
 `---

 ,-[ PoC for wrong remainder calculation in BPF_S_ANC_NLATTR_NEST ]--
 | ; (needs a fake netlink header at offset 0)
 | ld #0
 | ldx #42
 | ld #nlan
 | ret a
 `---

Fix the first issue by ensuring the message length fulfills the minimal
size constrains of a nla header. Fix the second bug by getting the math
for the remainder calculation right.

Fixes: 4738c1db15 ("[SKFILTER]: Add SKF_ADF_NLATTR instruction")
Fixes: d214c7537b ("filter: add SKF_AD_NLATTR_NEST to look for nested..")
Cc: Patrick McHardy <kaber@trash.net>
Cc: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv4: return valid RTA_IIF on ip route get
Julian Anastasov [Sun, 13 Apr 2014 15:08:02 +0000 (18:08 +0300)]
ipv4: return valid RTA_IIF on ip route get

[ Upstream commit 91146153da2feab18efab2e13b0945b6bb704ded ]

Extend commit 13378cad02afc2adc6c0e07fca03903c7ada0b37
("ipv4: Change rt->rt_iif encoding.") from 3.6 to return valid
RTA_IIF on 'ip route get ... iif DEVICE' instead of rt_iif 0
which is displayed as 'iif *'.

inet_iif is not appropriate to use because skb_iif is not set.
Use the skb->dev->ifindex instead.

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agovti: don't allow to add the same tunnel twice
Nicolas Dichtel [Fri, 11 Apr 2014 13:51:19 +0000 (15:51 +0200)]
vti: don't allow to add the same tunnel twice

[ Upstream commit 8d89dcdf80d88007647945a753821a06eb6cc5a5 ]

Before the patch, it was possible to add two times the same tunnel:
ip l a vti1 type vti remote 10.16.0.121 local 10.16.0.249 key 41
ip l a vti2 type vti remote 10.16.0.121 local 10.16.0.249 key 41

It was possible, because ip_tunnel_newlink() calls ip_tunnel_find() with the
argument dev->type, which was set only later (when calling ndo_init handler
in register_netdevice()). Let's set this type in the setup handler, which is
called before newlink handler.

Introduced by commit b9959fd3b0fa ("vti: switch to new ip tunnel code").

CC: Cong Wang <amwang@redhat.com>
CC: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agogre: don't allow to add the same tunnel twice
Nicolas Dichtel [Fri, 11 Apr 2014 13:51:18 +0000 (15:51 +0200)]
gre: don't allow to add the same tunnel twice

[ Upstream commit 5a4552752d8f7f4cef1d98775ece7adb7616fde2 ]

Before the patch, it was possible to add two times the same tunnel:
ip l a gre1 type gre remote 10.16.0.121 local 10.16.0.249
ip l a gre2 type gre remote 10.16.0.121 local 10.16.0.249

It was possible, because ip_tunnel_newlink() calls ip_tunnel_find() with the
argument dev->type, which was set only later (when calling ndo_init handler
in register_netdevice()). Let's set this type in the setup handler, which is
called before newlink handler.

Introduced by commit c54419321455 ("GRE: Refactor GRE tunneling code.").

CC: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv6: Limit mtu to 65575 bytes
Eric Dumazet [Fri, 11 Apr 2014 04:23:36 +0000 (21:23 -0700)]
ipv6: Limit mtu to 65575 bytes

[ Upstream commit 30f78d8ebf7f514801e71b88a10c948275168518 ]

Francois reported that setting big mtu on loopback device could prevent
tcp sessions making progress.

We do not support (yet ?) IPv6 Jumbograms and cook corrupted packets.

We must limit the IPv6 MTU to (65535 + 40) bytes in theory.

Tested:

ifconfig lo mtu 70000
netperf -H ::1

Before patch : Throughput :   0.05 Mbits

After patch : Throughput : 35484 Mbits

Reported-by: Francois WELLENREITER <f.wellenreiter@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>