KBL is similar to SKL
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

SKX is similar to BDX
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

This driver registers cpuidle devices when a CPU comes online, but it
leaves the registrations in place when a CPU goes offline.  The module
exit code only unregisters the currently online CPUs, leaving the
devices for offline CPUs dangling.

This patch changes the driver to clean up all registrations on exit,
even those from CPUs that are offline.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

If a cpuidle registration error occurs during the hot plug notifier
callback, we should really inform the hot plug machinery instead of
just ignoring the error.  This patch changes the callback to properly
return on error.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The helper function, intel_idle_cpu_init, registers one new device
with the cpuidle layer.  If the registration should fail, that
function immediately calls intel_idle_cpuidle_devices_uninit() to
unregister every last CPU's device.  However, it makes no sense to do
so, when called from the hot plug notifier callback.

This patch moves the call to intel_idle_cpuidle_devices_uninit()
outside of the helper function to the one call site that actually
needs to perform the de-registrations.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

This driver sets the broadcast tick quite early on during probe and does
not clean up again in cast of failure.  This patch moves the setup call
after the registration, placing the on_each_cpu() calls within the global
CPU lock region.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The helper function, intel_idle_cpuidle_devices_uninit, frees the
globally allocated per-CPU data.  However, this function is invoked
from the hot plug notifier callback at a time when freeing that data
is not safe.

If the call to cpuidle_register_driver() should fail (say, due to lack
of memory), then the driver will free its per-CPU region.  On the
*next* CPU_ONLINE event, the driver will happily use the region again
and even free it again if the failure repeats.

This patch fixes the issue by moving the call to free_percpu() outside
of the helper function at the two call sites that actually need to
free the per-CPU data.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

In the module_init() method, if the per-CPU allocation fails, then the
active cpuidle registration is not cleaned up.  This patch fixes the
issue by attempting the allocation before registration, and then
cleaning it up again on registration failure.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

In the module_exit() method, this driver first frees its per-CPU
pointer, then unregisters a callback making use of the pointer.
Furthermore, the function, intel_idle_cpuidle_devices_uninit, is racy
against CPU hot plugging as it calls for_each_online_cpu().

This patch corrects the issues by unregistering first on the exit path
while holding the hot plug lock.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The function, intel_idle_cpuidle_driver_init, makes calls on each CPU
to auto_demotion_disable() and c1e_promotion_disable().  These calls
are redundant, as intel_idle_cpu_init() does the same calls just a bit
later on.  They are also premature, as the driver registration may yet
fail.

This patch removes the redundant code.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The function, intel_idle_cpuidle_driver_init, delivers no error codes
at all.  This patch changes the function to return 'void' instead of
returning zero.
Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Signed-off-by: NRichard Cochran <rcochran@linutronix.de>
Signed-off-by: NLen Brown <len.brown@intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Enables "Intel(R) Xeon Phi(TM) Processor x200 Product Family" support,
formerly code-named KNL. It is based on modified Intel Atom Silvermont
microarchitecture.
Signed-off-by: NDasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
[micah.barany@intel.com: adjusted values of residency and latency]
Signed-off-by: NMicah Barany <micah.barany@intel.com>
[hubert.chrzaniuk@intel.com: removed deprecated CPUIDLE_FLAG_TIME_VALID flag]
Signed-off-by: NHubert Chrzaniuk <hubert.chrzaniuk@intel.com>
Signed-off-by: NPawel Karczewski <pawel.karczewski@intel.com>
Signed-off-by: NLen Brown <len.brown@intel.com>

Some SKL-H configurations require "intel_idle.max_cstate=7" to boot.
While that is an effective workaround, it disables C10.

This patch detects the problematic configuration,
and disables C8 and C9, keeping C10 enabled.

Note that enabling SGX in BIOS SETUP can also prevent this issue,
if the system BIOS provides that option.

https://bugzilla.kernel.org/show_bug.cgi?id=109081
"Freezes with Intel i7 6700HQ (Skylake), unless intel_idle.max_cstate=7"
Signed-off-by: NLen Brown <len.brown@intel.com>
Cc: stable@vger.kernel.org

Commit a9ceb78b (cpuidle,menu: use interactivity_req to disable
polling) changed the behavior of the fallback state selection part
of menu_select() so it looks at interactivity_req instead of
data->next_timer_us when it makes its decision.  That effectively
caused polling to be used more often as fallback idle which led to
significant increases of energy consumption in some cases.

Commit e132b9b3 (cpuidle: menu: use high confidence factors
only when considering polling) changed that logic again to be more
predictable, but that didn't help with the increased energy
consumption problem.

For this reason, go back to making decisions on which state to fall
back to based on data->next_timer_us which is the time we know for
sure something will happen rather than a prediction (which may be
inaccurate and turns out to be so often enough to be problematic).
However, take the target residency of the first proper idle state
(C1) into account, so that state is not used as the fallback one
if its target residency is greater than data->next_timer_us.

Fixes: a9ceb78b (cpuidle,menu: use interactivity_req to disable polling)
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Reported-and-tested-by: NDoug Smythies <dsmythies@telus.net>

The menu governor uses five different factors to pick the
idle state:
 - the user configured latency_req
 - the time until the next timer (next_timer_us)
 - the typical sleep interval, as measured recently
 - an estimate of sleep time by dividing next_timer_us by an observed factor
 - a load corrected version of the above, divided again by load

Only the first three items are known with enough confidence that
we can use them to consider polling, instead of an actual CPU
idle state, because the cost of being wrong about polling can be
excessive power use.

The latter two are used in the menu governor's main selection
loop, and can result in choosing a shallower idle state when
the system is expected to be busy again soon.

This pushes a busy system in the "performance" direction of
the performance<>power tradeoff, when choosing between idle
states, but stays more strictly on the "power" state when
deciding between polling and C1.
Signed-off-by: NRik van Riel <riel@redhat.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

We know that the avg variable actually ends up holding a 32 bit
quantity, since it's an average of such numbers. It is only a u64
because it is temporarily used to hold the sum. Making it an actual
u32 allows gcc to generate slightly better code, e.g. when computing
the square, it can do a 32x32->64 multiply.
Signed-off-by: NRasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Computing the integer square root is a rather expensive operation, at
least compared to doing a 64x64 -> 64 multiply (avg*avg) and, on 64
bit platforms, doing an extra comparison to a constant (variance <=
U64_MAX/36).

On 64 bit platforms, this does mean that we add a restriction on the
range of the variance where we end up using the estimate (since
previously the stddev <= ULONG_MAX was a tautology), but on the other
hand, we extend the range quite substantially on 32 bit platforms - in
both cases, we now allow standard deviations up to 715 seconds, which
is for example guaranteed if all observations are less than 1430
seconds.
Signed-off-by: NRasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Fix the RC QPs send queue overhead computation to take into account
two additional segments in the WQE which are needed for registration
operations.

The ATOMIC and UMR segments can't coexist together, so chose maximum out
of them.

The commit 9e65dc37 ("IB/mlx5: Fix RC transport send queue overhead
computation") was intended to update RC transport as commit messages
states, but added the code to UC transport.

Fixes: 9e65dc37 ("IB/mlx5: Fix RC transport send queue overhead computation")
Signed-off-by: NKamal Heib <kamalh@mellanox.com>
Signed-off-by: NLeon Romanovsky <leonro@mellanox.com>
Reviewed-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NDoug Ledford <dledford@redhat.com>

A narrow window for race condition still exist between
multicast join thread and *dev_flush workers.
A kernel crash caused by prolong erratic link state changes
was observed (most likely a faulty cabling):

[167275.656270] BUG: unable to handle kernel NULL pointer dereference at
0000000000000020
[167275.665973] IP: [<ffffffffa05f8f2e>] ipoib_mcast_join+0xae/0x1d0 [ib_ipoib]
[167275.674443] PGD 0
[167275.677373] Oops: 0000 [#1] SMP
...
[167275.977530] Call Trace:
[167275.982225]  [<ffffffffa05f92f0>] ? ipoib_mcast_free+0x200/0x200 [ib_ipoib]
[167275.992024]  [<ffffffffa05fa1b7>] ipoib_mcast_join_task+0x2a7/0x490
[ib_ipoib]
[167276.002149]  [<ffffffff8109d5fb>] process_one_work+0x17b/0x470
[167276.010754]  [<ffffffff8109e3cb>] worker_thread+0x11b/0x400
[167276.019088]  [<ffffffff8109e2b0>] ? rescuer_thread+0x400/0x400
[167276.027737]  [<ffffffff810a5aef>] kthread+0xcf/0xe0
Here was a hit spot:
ipoib_mcast_join() {
..............
      rec.qkey      = priv->broadcast->mcmember.qkey;
                                       ^^^^^^^
.....
 }
Proposed patch should prevent multicast join task to continue
if link state change is detected.
Signed-off-by: NAlex Estrin <alex.estrin@intel.com>

Changes from v4:
- as suggested by Doug Ledford, optimized spinlock usage,
i.e. ipoib_mcast_join() is called with lock held.
Changes from v3:
- sync with priv->lock before flag check.
Chages from v2:
- Move check for OPER_UP flag state to mcast_join() to
ensure no event worker is in progress.
- minor style fixes.
Changes from v1:
- No need to lock again if error detected.
Signed-off-by: NDoug Ledford <dledford@redhat.com>

When checking specific attribute from a bit mask, need to use bitwise
AND and not logical AND, fixed that.

Fixes: 145d9c54 ('IB/core: Display extended counter set if
available')
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Reviewed-by: NOr Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NDoug Ledford <dledford@redhat.com>

The while loop after err_slaves should use post-decrement; otherwise
we'll fail to do the kfrees for i==0, and will run into out-of-bounds
accesses if the setup above failed already at i==0.

[I'm not sure why one even bothers populating the ->vlan_filter array:
mlx4.h isn't #included by anything outside
drivers/net/ethernet/mellanox/mlx4/, and "git grep -C2 -w vlan_filter
drivers/net/ethernet/mellanox/mlx4/" seems to suggest that the
vlan_filter elements aren't used at all.]
Signed-off-by: NRasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: NDoug Ledford <dledford@redhat.com>

This reverts commit 829b6962.

Revert this change as it causes a sysfs path to change and therefore
introduces and ABI regression. More precisely Android's vold is not being
able to access /sys/module/mmcblk/parameters/perdev_minors any more, since
the path becomes changed to: "/sys/module/mmc_block/..."

Fixes: 829b6962 ("mmc: block: don't use parameter prefix if built as
module")
Reported-by: NJohn Stultz <john.stultz@linaro.org>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

In __cpufreq_cooling_register() we allocate the arrays for time_in_idle
and time_in_idle_timestamp to be as big as the number of cpus in this
cpufreq device.  However, in get_load() we access this array using the
cpu number as index, which can result in an out of bound access.

Index time_in_idle{,_timestamp} using the index in the cpufreq_device's
allowed_cpus mask, as we do for the load_cpu array in
cpufreq_get_requested_power()
Reported-by: NNicolas Boichat <drinkcat@chromium.org>
Cc: Amit Daniel Kachhap <amit.kachhap@gmail.com>
Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Eduardo Valentin <edubezval@gmail.com>
Tested-by: NNicolas Boichat <drinkcat@chromium.org>
Acked-by: NViresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: NJavi Merino <javi.merino@arm.com>
Signed-off-by: NEduardo Valentin <edubezval@gmail.com>

Intel BXT/APL use a card detect GPIO however the host controller
will not enable bus power unless it's card detect also reflects
the presence of a card.  Unfortunately those 2 things race which
can result in commands not starting, after which the controller
does nothing and there is a 10 second wait for the driver's
10-second timer to timeout.

That is fixed by having the driver look also at the present state
register to determine if the card is present.  Consequently, provide
a 'get_cd' mmc host operation for BXT/APL that does that.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org # v4.4+
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Intel BXT/APL use a card detect GPIO however the host controller
will not enable bus power unless it's card detect also reflects
the presence of a card.  Unfortunately those 2 things race which
can result in commands not starting, after which the controller
does nothing and there is a 10 second wait for the driver's
10-second timer to timeout.

That is fixed by having the driver look also at the present state
register to determine if the card is present.  Consequently, provide
a 'get_cd' mmc host operation for BXT/APL that does that.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org # v4.4+
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Drivers may need to provide their own get_cd() mmc host op, but
currently the internals of the current op (sdhci_get_cd()) are
provided by sdhci_do_get_cd() which is also called from
sdhci_request().

To allow override of the get_cd functionality, change sdhci_request()
to call ->get_cd() instead of sdhci_do_get_cd().

Note, in the future the call to ->get_cd() will likely be removed
from sdhci_request() since most drivers don't need actually it.
However this change is being done now to facilitate a subsequent
bug fix.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org # v4.4+
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

In the past, fixes for specific hardware devices were implemented
in sdhci using quirks.  That approach is no longer accepted because
the growing number of quirks was starting to make the code difficult
to understand and maintain.

One alternative to quirks, is to allow drivers to override the default
mmc host operations.  This patch makes it easy to do that, and it is
needed for a subsequent bug fix, for which separate patches are
provided.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org # v4.4+
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

EOImode1 is only used for the root controller and hence only the root
controller uses the eoimode1 functions for handling interrupts. However,
if the root controller supports EOImode1, then the EOImodeNS bit will be
set for all GICs, enabling EOImode1. This is not what we want and this
causes interrupts on non-root GICs to only be dropped in priority but
never deactivated. Therefore, only set the EOImodeNS bit for the root
controller.
Acked-by: NMarc Zyngier <marc.zyngier@arm.com>
Signed-off-by: NJon Hunter <jonathanh@nvidia.com>
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>

Setting the affinity of an IRQ, it only applicable for the root
interrupt controller and so only populate this operator for the root
controller.
Acked-by: NMarc Zyngier <marc.zyngier@arm.com>
Signed-off-by: NJon Hunter <jonathanh@nvidia.com>
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>

Calling component_add() may result in the completion of a set of
devices, which will try to bring up a master. In bringing the master
up, we populate its match array with the current set of children.

If binding any of the devices fails, component_add() itself will fail,
free the struct component entry, and return to the caller. The
now-freed entry is never removed from the master's match array, and
will later be used in a futile attempt to bind to freed memory.

Bring component_add's behaviour on failure to bring up a master into
line with component_del by removing the (to-be-freed) component from
the master's match array.

The specific case which broke was:
  - rockchip_drm_drv adds a component master
  - dwhdmi_rockchip adds a child component in probe (master incomplete)
  - rockchip_drm_vop adds two children in probe, which completes the
    set
  - inside component_add, we try to bring up the master, having
    populated the master's match array, and fail with EPROBE_DEFER from
    dwhdmi_rockchip; we delete the putative component
  - rockchip_drm_vop's probe fails and returns EPROBE_DEFER
  - we later re-probe rockchip_drm_vop and add the component; the
    master is complete, so we attempt to bring it up again
  - walking the match array, we find the previous child, whose master
    pointer doesn't match (as it has been freed in the meantime)
  - rockchip_drm_vop probe fails, and will never be attempted again

Fixes: ffc30b74Signed-off-by: NDaniel Stone <daniels@collabora.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Thierry Reding <treding@nvidia.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NBart Van Assche <bart.vanassche@sandisk.com>
Fixes: fb3269ba ("qla2xxx: Add selective command queuing")
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NBart Van Assche <bart.vanassche@sandisk.com>
Fixes: fb3269ba ("qla2xxx: Add selective command queuing")
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes an incorrect return of zero from the new
unmap_zeroes_data_store() configfs store attribute handler
introduced in v4.5-rc1, to use the correct 'count' bytes
return value.
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Otherwise we could try to evict overlapping userptr BOs in get_user_pages(),
leading to a possible circular locking dependency.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

An arbitrary amount of time can pass between spin_unlock and
radeon_fence_wait_any, so we need to ensure that nobody frees the
fences from under us.

Based on the analogous fix for amdgpu.
Signed-off-by: NNicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

Signed-off-by: NNicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

An arbitrary amount of time can pass between spin_unlock and
fence_wait_any_timeout, so we need to ensure that nobody frees the
fences from under us.

A stress test (rapidly starting and killing hundreds of glxgears
instances) ran into a deadlock in fence_wait_any_timeout after
about an hour, and this race condition appears to be a plausible
cause.

v2: agd: rebase on upstream
Signed-off-by: NNicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

No need to re-init asic if it's already been initialized.
Skip IB tests since kernel processes are frozen in thaw.
Signed-off-by: NFlora Cui <Flora.Cui@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Adding Intel codename DNV platform device IDs for SATA.
Signed-off-by: NAlexandra Yates <alexandra.yates@linux.intel.com>
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org