GitHub/moto-9609/android_kernel_motorola_exynos9610.git
8 years agocpufreq: mvebu: Move cpufreq code into drivers/cpufreq/
Viresh Kumar [Wed, 27 Apr 2016 03:22:27 +0000 (08:52 +0530)]
cpufreq: mvebu: Move cpufreq code into drivers/cpufreq/

Move cpufreq bits for mvebu into drivers/cpufreq/ directory, that's
where they really belong to.

Compiled tested only.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: dt: Kill platform-data
Viresh Kumar [Wed, 27 Apr 2016 03:22:26 +0000 (08:52 +0530)]
cpufreq: dt: Kill platform-data

There are no more users of platform-data for cpufreq-dt driver, get rid
of it.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agomvebu: Use dev_pm_opp_set_sharing_cpus() to mark OPP tables as shared
Viresh Kumar [Wed, 27 Apr 2016 03:22:25 +0000 (08:52 +0530)]
mvebu: Use dev_pm_opp_set_sharing_cpus() to mark OPP tables as shared

That will allow us to avoid using cpufreq-dt platform data.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: dt: Identify cpu-sharing for platforms without operating-points-v2
Viresh Kumar [Wed, 27 Apr 2016 03:22:24 +0000 (08:52 +0530)]
cpufreq: dt: Identify cpu-sharing for platforms without operating-points-v2

Existing platforms, which do not support operating-points-v2, can
explicitly tell the opp core that some of the CPUs share opp tables,
with help of dev_pm_opp_set_sharing_cpus().

For such platforms, explicitly ask the opp core to provide list of CPUs
sharing the opp table with current cpu device, before falling back to
platform data.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoMerge back earlier cpufreq material for v4.7.
Rafael J. Wysocki [Thu, 28 Apr 2016 13:19:31 +0000 (15:19 +0200)]
Merge back earlier cpufreq material for v4.7.

8 years agoPM / OPP: Add dev_pm_opp_get_sharing_cpus()
Viresh Kumar [Wed, 27 Apr 2016 03:22:23 +0000 (08:52 +0530)]
PM / OPP: Add dev_pm_opp_get_sharing_cpus()

OPP core allows a platform to mark OPP table as shared, when the
platform isn't using operating-points-v2 bindings.

And, so there should be a non DT way of finding out if the OPP table is
shared or not.

This patch adds dev_pm_opp_get_sharing_cpus(), which first tries to get
OPP sharing information from the opp-table (in case it is already marked
as shared), otherwise it uses the existing DT way of finding sharing
information.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: Mark cpumask as const in dev_pm_opp_set_sharing_cpus()
Viresh Kumar [Wed, 27 Apr 2016 03:22:22 +0000 (08:52 +0530)]
PM / OPP: Mark cpumask as const in dev_pm_opp_set_sharing_cpus()

dev_pm_opp_set_sharing_cpus() isn't supposed to update the cpumask
passed as its parameter, and so it should always have been marked
'const'.

Do it now.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: -ENOSYS is applicable only to syscalls
Viresh Kumar [Wed, 27 Apr 2016 03:22:21 +0000 (08:52 +0530)]
PM / OPP: -ENOSYS is applicable only to syscalls

Some of the routines have used -ENOSYS for the cases where the
functionality isn't implemented in the kernel. But ENOSYS is supposed to
be used only for syscalls.

Replace that with -ENOTSUPP, which specifically means that the operation
isn't supported.

While at it, replace exiting -EINVAL errors for similar cases to
-ENOTSUPP.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: governor: Change confusing struct field and variable names
Rafael J. Wysocki [Wed, 27 Apr 2016 23:19:03 +0000 (01:19 +0200)]
cpufreq: governor: Change confusing struct field and variable names

The name of the prev_cpu_wall field in struct cpu_dbs_info is
confusing, because it doesn't represent wall time, but the previous
update time as returned by get_cpu_idle_time() (that may be the
current value of jiffies_64 in some cases, for example).

Moreover, the names of some related variables in dbs_update() take
that confusion further.

Rename all of those things to make their names reflect the purpose
more accurately.  While at it, drop unnecessary parens from one of
the updated expressions.

No functional changes.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Chen Yu <yu.c.chen@intel.com>
8 years agocpufreq: intel_pstate: Enable PPC enforcement for servers
Srinivas Pandruvada [Wed, 27 Apr 2016 22:48:08 +0000 (15:48 -0700)]
cpufreq: intel_pstate: Enable PPC enforcement for servers

For platforms which are controlled via remove node manager, enable _PPC by
default. These platforms are mostly categorized as enterprise server or
performance servers. These platforms needs to go through some
certifications tests, which tests control via _PPC.
The relative risk of enabling by default is  low as this is is less likely
that these systems have broken _PSS table.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: intel_pstate: Adjust policy->max
Srinivas Pandruvada [Wed, 27 Apr 2016 22:48:07 +0000 (15:48 -0700)]
cpufreq: intel_pstate: Adjust policy->max

When policy->max is changed via _PPC or sysfs and is more than the max non
turbo frequency, it does not really change resulting performance in some
processors. When policy->max results in a P-State ratio more than the
turbo activation ratio, then processor can choose any P-State up to max
turbo. So the user or _PPC setting has no value, but this can cause
undesirable side effects like:
- Showing reduced max percentage in Intel P-State sysfs
- It can cause reduced max performance under certain boundary conditions:
The requested max scaling frequency either via _PPC or via cpufreq-sysfs,
will be converted into a fixed floating point max percent scale. In
majority of the cases this will result in correct max. But not 100% of the
time. If the _PPC is requested at a point where the calculation lead to a
lower max, this can result in a lower P-State then expected and it will
impact performance.
Example of this condition using a Broadwell laptop with config TDP.

ACPI _PSS table from a Broadwell laptop
2301000 2300000 2200000 2000000 1900000 1800000 1700000 1500000 1400000
1300000 1100000 1000000 900000 800000 600000 500000

The actual results by disabling config TDP so that we can get what is
requested on or below 2300000Khz.

scaling_max_freq        Max Requested P-State   Resultant scaling
max
---------------------------------------- ----------------------
2400000                 18                      2900000 (max
turbo)
2300000                 17                      2300000 (max
physical non turbo)
2200000                 15                      2100000
2100000                 15                      2100000
2000000                 13                      1900000
1900000                 13                      1900000
1800000                 12                      1800000
1700000                 11                      1700000
1600000                 10                      1600000
1500000                 f                       1500000
1400000                 e                       1400000
1300000                 d                       1300000
1200000                 c                       1200000
1100000                 a                       1000000
1000000                 a                       1000000
900000                  9                        900000
800000                  8                        800000
700000                  7                        700000
600000                  6                        600000
500000                  5                        500000
------------------------------------------------------------------

Now set the config TDP level 1 ratio as 0x0b (equivalent to 1100000KHz)
in BIOS (not every system will let you adjust this).
The turbo activation ratio will be set to one less than that, which will
be 0x0a (So any request above 1000000KHz should result in turbo region
assuming no thermal limits).
Here _PPC will request max to 1100000KHz (which basically should still
result in turbo as this is more than the turbo activation ratio up to
max allowable turbo frequency), but actual calculation resulted in a max
ceiling P-State which is 0x0a. So under any load condition, this driver
will not request turbo P-States. This will be a huge performance hit.

When config TDP feature is ON, if the _PPC points to a frequency above
turbo activation ratio, the performance can still reach max turbo. In this
case we don't need to treat this as the reduced frequency in set_policy
callback.

In this change when config TDP is active (by checking if the physical max
non turbo ratio is more than the current max non turbo ratio), any request
above current max non turbo is treated as full performance.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw : Minor cleanups ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: intel_pstate: Enforce _PPC limits
Srinivas Pandruvada [Wed, 27 Apr 2016 22:48:06 +0000 (15:48 -0700)]
cpufreq: intel_pstate: Enforce _PPC limits

Use ACPI _PPC notification to limit max P state driver will request.
ACPI _PPC change notification is sent by BIOS to limit max P state
in several cases:
- Reduce impact of platform thermal condition
- When Config TDP feature is used, a changed _PPC is sent to
follow TDP change
- Remote node managers in server want to control platform power
via baseboard management controller (BMC)

This change registers with ACPI processor performance lib so that
_PPC changes are notified to cpufreq core, which in turns will
result in call to .setpolicy() callback. Also the way _PSS
table identifies a turbo frequency is not compatible to max turbo
frequency in intel_pstate, so the very first entry in _PSS needs
to be adjusted.

This feature can be turned on by using kernel parameters:
intel_pstate=support_acpi_ppc

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw: Minor cleanups ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: powernv: Ramp-down global pstate slower than local-pstate
Akshay Adiga [Tue, 19 Apr 2016 09:58:01 +0000 (15:28 +0530)]
cpufreq: powernv: Ramp-down global pstate slower than local-pstate

The frequency transition latency from pmin to pmax is observed to be in
few millisecond granurality. And it usually happens to take a performance
penalty during sudden frequency rampup requests.

This patch set solves this problem by using an entity called "global
pstates". The global pstate is a Chip-level entity, so the global entitiy
(Voltage) is managed across the cores. The local pstate is a Core-level
entity, so the local entity (frequency) is managed across threads.

This patch brings down global pstate at a slower rate than the local
pstate. Hence by holding global pstates higher than local pstate makes
the subsequent rampups faster.

A per policy structure is maintained to keep track of the global and
local pstate changes. The global pstate is brought down using a parabolic
equation. The ramp down time to pmin is set to ~5 seconds. To make sure
that the global pstates are dropped at regular interval , a timer is
queued for every 2 seconds during ramp-down phase, which eventually brings
the pstate down to local pstate.

Iozone results show fairly consistent performance boost.
YCSB on redis shows improved Max latencies in most cases.

Iozone write/rewite test were made with filesizes 200704Kb and 401408Kb
with different record sizes . The following table shows IOoperations/sec
with and without patch.

Iozone Results ( in op/sec) ( mean over 3 iterations )
---------------------------------------------------------------------
file size-                      with            without   %
recordsize-IOtype               patch           patch change
----------------------------------------------------------------------
200704-1-SeqWrite               1616532         1615425         0.06
200704-1-Rewrite                2423195         2303130         5.21
200704-2-SeqWrite               1628577         1602620         1.61
200704-2-Rewrite                2428264         2312154         5.02
200704-4-SeqWrite               1617605         1617182         0.02
200704-4-Rewrite                2430524         2351238         3.37
200704-8-SeqWrite               1629478         1600436         1.81
200704-8-Rewrite                2415308         2298136         5.09
200704-16-SeqWrite              1619632         1618250         0.08
200704-16-Rewrite               2396650         2352591         1.87
200704-32-SeqWrite              1632544         1598083         2.15
200704-32-Rewrite               2425119         2329743         4.09
200704-64-SeqWrite              1617812         1617235         0.03
200704-64-Rewrite               2402021         2321080         3.48
200704-128-SeqWrite             1631998         1600256         1.98
200704-128-Rewrite              2422389         2304954         5.09
200704-256 SeqWrite             1617065         1616962         0.00
200704-256-Rewrite              2432539         2301980         5.67
200704-512-SeqWrite             1632599         1598656         2.12
200704-512-Rewrite              2429270         2323676         4.54
200704-1024-SeqWrite            1618758         1616156         0.16
200704-1024-Rewrite             2431631         2315889         4.99
401408-1-SeqWrite               1631479         1608132         1.45
401408-1-Rewrite                2501550         2459409         1.71
401408-2-SeqWrite               1617095         1626069         -0.55
401408-2-Rewrite                2507557         2443621         2.61
401408-4-SeqWrite               1629601         1611869         1.10
401408-4-Rewrite                2505909         2462098         1.77
401408-8-SeqWrite               1617110         1626968         -0.60
401408-8-Rewrite                2512244         2456827         2.25
401408-16-SeqWrite              1632609         1609603         1.42
401408-16-Rewrite               2500792         2451405         2.01
401408-32-SeqWrite              1619294         1628167         -0.54
401408-32-Rewrite               2510115         2451292         2.39
401408-64-SeqWrite              1632709         1603746         1.80
401408-64-Rewrite               2506692         2433186         3.02
401408-128-SeqWrite             1619284         1627461         -0.50
401408-128-Rewrite              2518698         2453361         2.66
401408-256-SeqWrite             1634022         1610681         1.44
401408-256-Rewrite              2509987         2446328         2.60
401408-512-SeqWrite             1617524         1628016         -0.64
401408-512-Rewrite              2504409         2442899         2.51
401408-1024-SeqWrite            1629812         1611566         1.13
401408-1024-Rewrite             2507620          2442968        2.64

Tested with YCSB workload (50% update + 50% read) over redis for 1 million
records and 1 million operation. Each test was carried out with target
operations per second and persistence disabled.

Max-latency (in us)( mean over 5 iterations )
---------------------------------------------------------------
op/s    Operation       with patch      without patch   %change
---------------------------------------------------------------
15000   Read            61480.6         50261.4         22.32
15000   cleanup         215.2           293.6           -26.70
15000   update          25666.2         25163.8         2.00

25000   Read            32626.2         89525.4         -63.56
25000   cleanup         292.2           263.0           11.10
25000   update          32293.4         90255.0         -64.22

35000   Read            34783.0         33119.0         5.02
35000   cleanup         321.2           395.8           -18.8
35000   update          36047.0         38747.8         -6.97

40000   Read            38562.2         42357.4         -8.96
40000   cleanup         371.8           384.6           -3.33
40000   update          27861.4         41547.8         -32.94

45000   Read            42271.0         88120.6         -52.03
45000   cleanup         263.6           383.0           -31.17
45000   update          29755.8         81359.0         -63.43

(test without target op/s)
47659   Read            83061.4         136440.6        -39.12
47659   cleanup         195.8           193.8           1.03
47659   update          73429.4         124971.8        -41.24

Signed-off-by: Akshay Adiga <akshay.adiga@linux.vnet.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: powernv: Remove flag use-case of policy->driver_data
Shilpasri G Bhat [Tue, 19 Apr 2016 09:58:00 +0000 (15:28 +0530)]
cpufreq: powernv: Remove flag use-case of policy->driver_data

commit 1b0289848d5d ("cpufreq: powernv: Add sysfs attributes to show
throttle stats") used policy->driver_data as a flag for one-time creation
of throttle sysfs files. Instead of this use 'kernfs_find_and_get()' to
check if the attribute already exists. This is required as
policy->driver_data is used for other purposes in the later patch.

Signed-off-by: Shilpasri G Bhat <shilpa.bhat@linux.vnet.ibm.com>
Signed-off-by: Akshay Adiga <akshay.adiga@linux.vnet.ibm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: e_powersaver: Use IS_ENABLED() instead of checking for built-in or module
Javier Martinez Canillas [Wed, 27 Apr 2016 00:14:13 +0000 (20:14 -0400)]
cpufreq: e_powersaver: Use IS_ENABLED() instead of checking for built-in or module

The IS_ENABLED() macro checks if a Kconfig symbol has been enabled either
built-in or as a module, use that macro instead of open coding the same.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: governor: Fix prev_load initialization in cpufreq_governor_start()
Rafael J. Wysocki [Mon, 25 Apr 2016 14:21:34 +0000 (16:21 +0200)]
cpufreq: governor: Fix prev_load initialization in cpufreq_governor_start()

The way cpufreq_governor_start() initializes j_cdbs->prev_load is
questionable.

First off, j_cdbs->prev_cpu_wall used as a denominator in the
computation may be zero.  The case this happens is when
get_cpu_idle_time_us() returns -1 and get_cpu_idle_time_jiffy()
used to return that number is called exactly at the jiffies_64
wrap time.  It is rather hard to trigger that error, but it is not
impossible and it will just crash the kernel then.

Second, j_cdbs->prev_load is computed as the average load during
the entire time since the system started and it may not reflect the
load in the previous sampling period (as it is expected to).
That doesn't play well with the way dbs_update() uses that value.
Namely, if the update time delta (wall_time) happens do be greater
than twice the sampling rate on the first invocation of it, the
initial value of j_cdbs->prev_load (which may be completely off) will
be returned to the caller as the current load (unless it is equal to
zero and unless another CPU sharing the same policy object has a
greater load value).

For this reason, notice that the prev_load field of struct cpu_dbs_info
is only used by dbs_update() and only in that one place, so if
cpufreq_governor_start() is modified to always initialize it to 0,
it will make dbs_update() always compute the actual load first time
it checks the update time delta against the doubled sampling rate
(after initialization) and there won't be any side effects of it.

Consequently, modify cpufreq_governor_start() as described.

Fixes: 18b46abd0009 (cpufreq: governor: Be friendly towards latency-sensitive bursty workloads)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
8 years agocpufreq: hisilicon: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:47 +0000 (16:58 +0530)]
cpufreq: hisilicon: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: zynq: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:46 +0000 (16:58 +0530)]
cpufreq: zynq: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: sunxi: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:45 +0000 (16:58 +0530)]
cpufreq: sunxi: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: shmobile: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:44 +0000 (16:58 +0530)]
cpufreq: shmobile: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: rockchip: Use generic platdev driver
Finley Xiao [Fri, 22 Apr 2016 11:28:43 +0000 (16:58 +0530)]
cpufreq: rockchip: Use generic platdev driver

This patch add rockchip's compatible string to the compat list and
remove similar code from platform code for supporting generic platdev
driver.

Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: omap: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:42 +0000 (16:58 +0530)]
cpufreq: omap: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: imx: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:41 +0000 (16:58 +0530)]
cpufreq: imx: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Note that the complete routine imx27_dt_init() is removed as

of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);

has same effect as a NULL .init_machine machine callback pointer.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Lucas Stach <l.stach@pengutronix.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: berlin: Use generic platdev driver
Viresh Kumar [Fri, 22 Apr 2016 11:28:40 +0000 (16:58 +0530)]
cpufreq: berlin: Use generic platdev driver

The cpufreq-dt-platdev driver supports creation of cpufreq-dt platform
device now, reuse that and remove similar code from platform code.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: dt: Mark platdev machines array as __initconst
Viresh Kumar [Fri, 22 Apr 2016 11:28:39 +0000 (16:58 +0530)]
cpufreq: dt: Mark platdev machines array as __initconst

The machines array in cpufreq-dt-platdev is used only once at boot time
and so should be marked with __initconst, so that kernel can free up
memory used for it, if required.

Suggested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: Mark shared-opp for non-dt case
Viresh Kumar [Thu, 21 Apr 2016 08:58:57 +0000 (14:28 +0530)]
PM / OPP: Mark shared-opp for non-dt case

opp core allows OPPs to be explicitly marked as shared from platform
code, in case of operating-point v1 bindings.

Though we do everything fine in that case, we don't set the flag in the
opp-table to indicate that the OPPs are shared. It works fine today as
the flag isn't used anywhere else in the core, but we should be doing
the right thing by marking it set.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: Relocate dev_pm_opp_set_sharing_cpus()
Viresh Kumar [Thu, 21 Apr 2016 08:58:56 +0000 (14:28 +0530)]
PM / OPP: Relocate dev_pm_opp_set_sharing_cpus()

Move dev_pm_opp_set_sharing_cpus() towards the end of the file. This
is required for better readability after the next patch is applied,
which adds dev_pm_opp_get_sharing_cpus().

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: dev_pm_opp_set_sharing_cpus() doesn't depend on CONFIG_OF
Viresh Kumar [Thu, 21 Apr 2016 08:58:55 +0000 (14:28 +0530)]
PM / OPP: dev_pm_opp_set_sharing_cpus() doesn't depend on CONFIG_OF

dev_pm_opp_set_sharing_cpus() doesn't do any DT specific stuff and its
declarations are added within the CONFIG_OF ifdef by mistake. Take them
out of that.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: Add missing doc style comments
Viresh Kumar [Thu, 21 Apr 2016 08:58:54 +0000 (14:28 +0530)]
PM / OPP: Add missing doc style comments

Few of the routines in cpu.c were missing these, add them.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoPM / OPP: Propagate the error returned by _find_opp_table()
Viresh Kumar [Thu, 21 Apr 2016 08:58:53 +0000 (14:28 +0530)]
PM / OPP: Propagate the error returned by _find_opp_table()

Don't send -EINVAL and propagate what's received from _find_opp_table().

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: qoriq: Fix cooling device registration issue during suspend
Jia Hongtao [Tue, 19 Apr 2016 09:00:07 +0000 (17:00 +0800)]
cpufreq: qoriq: Fix cooling device registration issue during suspend

Cooling device is registered by ready callback. It's also invoked while
system resuming from sleep (Enabling non-boot cpus). Thus cooling device
may be multiple registered. Matchable unregistration is added to exit
callback to fix this issue.

Signed-off-by: Jia Hongtao <hongtao.jia@nxp.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: qoriq: Remove __exit macro from .exit callback
Jia Hongtao [Tue, 19 Apr 2016 09:00:06 +0000 (17:00 +0800)]
cpufreq: qoriq: Remove __exit macro from .exit callback

.exit callback (qoriq_cpufreq_cpu_exit()) is also used during suspend.
So __exit macro should be removed or the function will be discarded.

Signed-off-by: Jia Hongtao <hongtao.jia@nxp.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: qoriq: Don't show cooling device messages if THERMAL_OF undefined
Jia Hongtao [Mon, 18 Apr 2016 07:59:32 +0000 (15:59 +0800)]
cpufreq: qoriq: Don't show cooling device messages if THERMAL_OF undefined

When THERMAL_OF is undefined the cooling device messages should not be
shown. -ENOSYS is returned from of_cpufreq_cooling_register() when
THERMAL_OF is undefined.

Signed-off-by: Jia Hongtao <hongtao.jia@nxp.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: ACPI / CPPC: Add module support for cppc_cpufreq driver
Ashwin Chaugule [Fri, 15 Apr 2016 00:45:53 +0000 (20:45 -0400)]
cpufreq: ACPI / CPPC: Add module support for cppc_cpufreq driver

Add a function to cleanup at module exit and export
appropriate GPL string to enable moduler support
for the cppc_cpufreq driver.

Reported-by: Srinivas Pandruvada <srinivas.pandruvada@intel.com>
Signed-off-by: Ashwin Chaugule <ashwin.chaugule@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agocpufreq: intel_pstate: Use average P-State instead of current P-State
Philippe Longepe [Fri, 22 Apr 2016 18:46:09 +0000 (11:46 -0700)]
cpufreq: intel_pstate: Use average P-State instead of current P-State

The result returned by pid_calc() is subtracted from current_pstate
(which is the P-State requested during the last period) in order to
obtain the target P-State for the current iteration.

However, current_pstate may not reflect the real current P-State of
the CPU. In particular, that P-State may be higher because of the
frequency sharing per module.

The theory is:
 - The load is the percentage of time spent in C0 and is related to
   the average P-State during the same period.
 - The last requested P-State can be completely different than the
   average P-State (because of frequency sharing or throttling).
 - The P-State shift computed by the pid_calc is based on the load
   computed at average P-State, so the shift must be relative to
   this average P-State.

Using the average P-State instead of current P-State improves power
without significant performance penalty in cases when a task migrates
from one core to other core sharing frequency and voltage.

Performance and power comparison with this patch on Cherry Trail
platform using Android:

Benchmark               ?Perf    ?Power
FishTank                10.45%    3.1%
SmartBench-Gaming       -0.1%   -10.4%
SmartBench-Productivity -0.8%   -10.4%
CandyCrush                n/a   -17.4%
AngryBirds                n/a    -5.9%
videoPlayback             n/a   -13.9%
audioPlayback             n/a    -4.9%
IcyRocks-20-50           0.0%   -38.4%
iozone RR               -0.16%  -1.3%
iozone RW                0.74%  -1.3%

Signed-off-by: Philippe Longepe <philippe.longepe@linux.intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
8 years agoMerge back cpufreq changes for v4.7.
Rafael J. Wysocki [Mon, 25 Apr 2016 13:44:01 +0000 (15:44 +0200)]
Merge back cpufreq changes for v4.7.

8 years agoRevert "cpufreq: governor: Fix negative idle_time when configured with CONFIG_HZ_PERI...
Rafael J. Wysocki [Thu, 21 Apr 2016 18:57:47 +0000 (20:57 +0200)]
Revert "cpufreq: governor: Fix negative idle_time when configured with CONFIG_HZ_PERIODIC"

Revert commit 0df35026c6a5 (cpufreq: governor: Fix negative idle_time
when configured with CONFIG_HZ_PERIODIC) that introduced a regression
by causing the ondemand cpufreq governor to misbehave for
CONFIG_TICK_CPU_ACCOUNTING unset (the frequency goes up to the max at
one point and stays there indefinitely).

The revert takes subsequent modifications of the code in question into
account.

Fixes: 0df35026c6a5 (cpufreq: governor: Fix negative idle_time when configured with CONFIG_HZ_PERIODIC)
Link: https://bugzilla.kernel.org/show_bug.cgi?id=115261
Reported-and-tested-by: Timo Valtoaho <timo.valtoaho@gmail.com>
Cc: 4.5+ <stable@vger.kernel.org> # 4.5+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
8 years agoLinux 4.6-rc5
Linus Torvalds [Sun, 24 Apr 2016 23:17:05 +0000 (16:17 -0700)]
Linux 4.6-rc5

8 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux...
Linus Torvalds [Sun, 24 Apr 2016 00:15:39 +0000 (17:15 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/evalenti/linux-soc-thermal

Pull thermal fixes from Eduardo Valentin:
 "Specifics in this pull request:

   - Fixes in mediatek and OF thermal drivers

   - Fixes in power_allocator governor

   - More fixes of unsigned to int type change in thermal_core.c.

  These change have been CI tested using KernelCI bot. \o/"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal:
  thermal: fix Mediatek thermal controller build
  thermal: consistently use int for trip temp
  thermal: fix mtk_thermal build dependency
  thermal: minor mtk_thermal.c cleanups
  thermal: power_allocator: req_range multiplication should be a 64 bit type
  thermal: of: add __init attribute

8 years agoMerge tag 'asm-generic-4.6' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd...
Linus Torvalds [Sat, 23 Apr 2016 21:53:11 +0000 (14:53 -0700)]
Merge tag 'asm-generic-4.6' of git://git./linux/kernel/git/arnd/asm-generic

Pull asm-generic update from Arnd Bergmann:
 "Here is one patch to wire up the preadv/pwritev system calls in the
  generic system call table, which is required for all architectures
  that were merged in the last few years, including arm64.

  Usually these get merged along with the syscall implementation or one
  of the architecture trees, but this time that did not happen.

  Andre and Christoph both sent a version of this patch, I picked the
  one I got first"

* tag 'asm-generic-4.6' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  generic syscalls: wire up preadv2 and pwritev2 syscalls

8 years agogeneric syscalls: wire up preadv2 and pwritev2 syscalls
Andre Przywara [Mon, 11 Apr 2016 09:17:46 +0000 (10:17 +0100)]
generic syscalls: wire up preadv2 and pwritev2 syscalls

These new syscalls are implemented as generic code, so enable them for
architectures like arm64 which use the generic syscall table.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
8 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 23 Apr 2016 19:07:29 +0000 (12:07 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Ingo Molnar:
 "Misc fixes: two EDAC driver fixes, a Xen crash fix, a HyperV log spam
  fix and a documentation fix"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86 EDAC, sb_edac.c: Take account of channel hashing when needed
  x86 EDAC, sb_edac.c: Repair damage introduced when "fixing" channel address
  x86/mm/xen: Suppress hugetlbfs in PV guests
  x86/doc: Correct limits in Documentation/x86/x86_64/mm.txt
  x86/hyperv: Avoid reporting bogus NMI status for Gen2 instances

8 years agoMerge branches 'perf-urgent-for-linus', 'smp-urgent-for-linus' and 'timers-urgent...
Linus Torvalds [Sat, 23 Apr 2016 18:45:52 +0000 (11:45 -0700)]
Merge branches 'perf-urgent-for-linus', 'smp-urgent-for-linus' and 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf, cpu hotplug and timer fixes from Ingo Molnar:
 "perf:
   - A single tooling fix for a user-triggerable segfault.

  CPU hotplug:
   - Fix a CPU hotplug corner case regression, introduced by the recent
     hotplug rework

  timers:
   - Fix a boot hang in the ARM based Tango SoC clocksource driver"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf intel-pt: Fix segfault tracing transactions

* 'smp-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  cpu/hotplug: Fix rollback during error-out in __cpu_disable()

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource/drivers/tango-xtal: Fix boot hang due to incorrect test

8 years agoMerge branch 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 23 Apr 2016 18:39:48 +0000 (11:39 -0700)]
Merge branch 'locking-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull locking fixes from Ingo Molnar:
 "Misc fixes:

  pvqspinlocks:
   - an instrumentation fix

  futexes:
   - preempt-count vs pagefault_disable decouple corner case fix
   - futex requeue plist race window fix
   - futex UNLOCK_PI transaction fix for a corner case"

* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  asm-generic/futex: Re-enable preemption in futex_atomic_cmpxchg_inatomic()
  futex: Acknowledge a new waiter in counter before plist
  futex: Handle unlock_pi race gracefully
  locking/pvqspinlock: Fix division by zero in qstat_read()

8 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 23 Apr 2016 18:34:39 +0000 (11:34 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Ingo Molnar:
 "A core irq affinity masks related fix and a MIPS irqchip driver fix"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/mips-gic: Don't overrun pcpu_masks array
  genirq: Dont allow affinity mask to be updated on IPIs

8 years agoMerge branch 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 23 Apr 2016 18:25:01 +0000 (11:25 -0700)]
Merge branch 'core-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull objtool fixes from Ingo Molnar:
 "A handful of objtool fixes: two improvements to how warnings are
  printed plus a false positive warning fix, and build environment fix"

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  objtool: Fix Makefile to properly see if libelf is supported
  objtool: Detect falling through to the next function
  objtool: Add workaround for GCC switch jump table bug

8 years agoMerge tag 'usb-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sat, 23 Apr 2016 18:20:03 +0000 (11:20 -0700)]
Merge tag 'usb-4.6-rc5' of git://git./linux/kernel/git/gregkh/usb

Pull USB / PHY driver fixes from Greg KH:
 "Here are two small sets of patches, both from subsystem trees, USB
  gadget and PHY drivers.

  Full details are in the shortlog, and they have all been in linux-next
  for a while (before I merged them to the USB tree)"

* tag 'usb-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: gadget: f_fs: Fix use-after-free
  usb: dwc3: gadget: Fix suspend/resume during device mode
  usb: dwc3: fix memory leak of dwc->regset
  usb: dwc3: core: fix PHY handling during suspend
  usb: dwc3: omap: fix up error path on probe()
  usb: gadget: composite: Clear reserved fields of SSP Dev Cap
  phy: rockchip-emmc: adapt binding to specifiy register offset and length
  phy: rockchip-emmc: should be a child device of the GRF
  phy: rockchip-dp: should be a child device of the GRF

8 years agoMerge tag 'tty-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sat, 23 Apr 2016 18:13:46 +0000 (11:13 -0700)]
Merge tag 'tty-4.6-rc5' of git://git./linux/kernel/git/gregkh/tty

Pull serial fixes from Greg KH:
 "Here are 3 serial driver fixes for issues that have been reported.

  Two are reverts, fixing problems that were in the big TTY/Serial
  driver merge in 4.6-rc1, and the last one is a simple bugfix for a
  regression that showed up in 4.6-rc1 as well.

  All have been in linux-next with no reported issues"

* tag 'tty-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  Revert "serial: 8250: Add hardware dependency to RT288X option"
  tty/serial/8250: fix RS485 half-duplex RX
  Revert "serial-uartlite: Constify uartlite_be/uartlite_le"

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sat, 23 Apr 2016 18:04:26 +0000 (11:04 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input subsystem fixes from Dmitry Torokhov:
 "Just minor driver fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: twl4030-vibra - do not reparent to grandparent
  Input: twl6040-vibra - do not reparent to grandparent
  Input: twl6040-vibra - ignore return value of schedule_work
  Input: twl6040-vibra - fix NULL pointer dereference by removing workqueue
  Input: pmic8xxx-pwrkey - fix algorithm for converting trigger delay
  Input: arizona-haptic - don't assign input_dev parent
  Input: clarify we want BTN_TOOL_<name> on proximity
  Input: xpad - add Mad Catz FightStick TE 2 VID/PID
  Input: gtco - fix crash on detecting device without endpoints

8 years agoMerge tag 'pinctrl-v4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Fri, 22 Apr 2016 18:52:49 +0000 (11:52 -0700)]
Merge tag 'pinctrl-v4.6-3' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:
 "Some pin control driver fixes came in.  One headed for stable and the
  other two are just ordinary merge window fixes.

   - Make the i.MX driver select REGMAP as a dependency
   - Fix up the Mediatek debounce time unit
   - Fix a real hairy ffs vs __ffs issue in the Single pinctrl driver"

* tag 'pinctrl-v4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: single: Fix pcs_parse_bits_in_pinctrl_entry to use __ffs than ffs
  pinctrl: mediatek: correct debounce time unit in mtk_gpio_set_debounce
  pinctrl: imx: Kconfig: PINCTRL_IMX select REGMAP

8 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 22 Apr 2016 18:11:15 +0000 (11:11 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Catalin Marinas:

 - Cache invalidation fix for early CPU boot status update (incorrect
   cacheline)

 - of_put_node() missing in the spin_table code

 - EL1/El2 early init inconsistency when Virtualisation Host Extensions
   are present

 - RCU warning fix in the arm_pmu.c driver

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Fix EL1/EL2 early init inconsistencies with VHE
  drivers/perf: arm-pmu: fix RCU usage on pmu resume from low-power
  arm64: spin-table: add missing of_node_put()
  arm64: fix invalidation of wrong __early_cpu_boot_status cacheline

8 years agoMerge tag 'powerpc-4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Fri, 22 Apr 2016 17:53:12 +0000 (10:53 -0700)]
Merge tag 'powerpc-4.6-3' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:
 "Three powerpc cpu feature fixes from Anton Blanchard:

   - scan_features() updated incorrect bits for REAL_LE

   - update cpu_user_features2 in scan_features()

   - update TM user feature bits in scan_features()"

* tag 'powerpc-4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc: Update TM user feature bits in scan_features()
  powerpc: Update cpu_user_features2 in scan_features()
  powerpc: scan_features() updates incorrect bits for REAL_LE

8 years agoMerge tag 'iommu-fixes-v4.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 22 Apr 2016 17:41:31 +0000 (10:41 -0700)]
Merge tag 'iommu-fixes-v4.6-rc4' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU fixes from Joerg Roedel:
 "The fixes include:

   - Two patches to revert the use of default domains in the ARM SMMU
     driver.  Enabling this caused regressions which need more thorough
     fixing.  So the regressions are fixed for now by disabling the use
     of default domains.

   - A fix for a v4.4 regression in the AMD IOMMU driver which broke
     devices behind invisible PCIe-to-PCI bridges with IOMMU enabled"

* tag 'iommu-fixes-v4.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/arm-smmu: Don't allocate resources for bypass domains
  iommu/arm-smmu: Fix stream-match conflict with IOMMU_DOMAIN_DMA
  iommu/amd: Fix checking of pci dma aliases

8 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 22 Apr 2016 17:29:52 +0000 (10:29 -0700)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "i915, nouveau and amdgpu/radeon fixes in this:

  nouveau:
     Two fixes, one for a regression with dithering and one for a bug
     hit by the userspace drivers.

  i915:
     A few fixes, mostly things heading for stable, two important
     skylake GT3/4 hangs.

  radeon/amdgpu:
     Some audio, suspend/resume and some runtime PM fixes, along with
     two patches to harden the userptr ABI a bit"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (24 commits)
  drm: Loongson-3 doesn't fully support wc memory
  drm/nouveau/gr/gf100: select a stream master to fixup tfb offset queries
  amdgpu/uvd: add uvd fw version for amdgpu
  drm/amdgpu: forbid mapping of userptr bo through radeon device file
  drm/radeon: forbid mapping of userptr bo through radeon device file
  drm/amdgpu: bump the afmt limit for CZ, ST, Polaris
  drm/amdgpu: use defines for CRTCs and AMFT blocks
  drm/dp/mst: Validate port in drm_dp_payload_send_msg()
  drm/nouveau/kms: fix setting of default values for dithering properties
  drm/radeon: print a message if ATPX dGPU power control is missing
  Revert "drm/radeon: disable runtime pm on PX laptops without dGPU power control"
  drm/amdgpu/acp: fix resume on CZ systems with AZ audio
  drm/radeon: add a quirk for a XFX R9 270X
  drm/radeon: print pci revision as well as pci ids on driver load
  drm/i915: Use fw_domains_put_with_fifo() on HSW
  drm/i915: Force ringbuffers to not be at offset 0
  drm/i915: Adjust size of PIPE_CONTROL used for gen8 render seqno write
  drm/i915/skl: Fix spurious gpu hang with gt3/gt4 revs
  drm/i915/skl: Fix rc6 based gpu/system hang
  drm/i915/userptr: Hold mmref whilst calling get-user-pages
  ...

8 years agoMerge tag 'sound-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 22 Apr 2016 17:17:18 +0000 (10:17 -0700)]
Merge tag 'sound-4.6-rc5' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Again a relatively calm week without surprise: most of fixes are about
  HD-audio, including fixes for Cirrus codec regression and a race over
  regmap access.  Although both change are slightly unintuitive, the
  risk of further breakage is quite low, I hope.

  Other than that, all the rest are trivial"

* tag 'sound-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda - Fix possible race on regmap bypass flip
  ALSA: pcxhr: Fix missing mutex unlock
  ALSA: hda - add PCI ID for Intel Broxton-T
  ALSA: hda - Keep powering up ADCs on Cirrus codecs
  ALSA: hda/realtek - Add ALC3234 headset mode for Optiplex 9020m
  ALSA - hda: hdmi check NULL pointer in hdmi_set_chmap
  ALSA: hda - Don't trust the reported actual power state

8 years agoMerge tag 'phy-for-4.6-rc' of git://git.kernel.org/pub/scm/linux/kernel/git/kishon...
Greg Kroah-Hartman [Fri, 22 Apr 2016 08:13:24 +0000 (17:13 +0900)]
Merge tag 'phy-for-4.6-rc' of git://git./linux/kernel/git/kishon/linux-phy into usb-linus

Kishon writes:

phy: for 4.6-rc

*) make rockchip-dp and rockchip-emmc PHY child device of
   GRF

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
8 years agox86 EDAC, sb_edac.c: Take account of channel hashing when needed
Tony Luck [Thu, 14 Apr 2016 17:22:02 +0000 (10:22 -0700)]
x86 EDAC, sb_edac.c: Take account of channel hashing when needed

Haswell and Broadwell can be configured to hash the channel
interleave function using bits [27:12] of the physical address.

On those processor models we must check to see if hashing is
enabled (bit21 of the HASWELL_HASYSDEFEATURE2 register) and
act accordingly.

Based on a patch by patrickg <patrickg@supermicro.com>

Tested-by: Patrick Geary <patrickg@supermicro.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Acked-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Cc: Aristeu Rozanski <arozansk@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-edac@vger.kernel.org
Cc: stable@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
8 years agox86 EDAC, sb_edac.c: Repair damage introduced when "fixing" channel address
Tony Luck [Thu, 14 Apr 2016 17:21:52 +0000 (10:21 -0700)]
x86 EDAC, sb_edac.c: Repair damage introduced when "fixing" channel address

In commit:

  eb1af3b71f9d ("Fix computation of channel address")

I switched the "sck_way" variable from holding the log2 value read
from the h/w to instead be the actual number. Unfortunately it
is needed in log2 form when used to shift the address.

Tested-by: Patrick Geary <patrickg@supermicro.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Acked-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Cc: Aristeu Rozanski <arozansk@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-edac@vger.kernel.org
Cc: stable@vger.kernel.org
Fixes: eb1af3b71f9d ("Fix computation of channel address")
Signed-off-by: Ingo Molnar <mingo@kernel.org>
8 years agoMerge tag 'fixes-for-v4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi...
Greg Kroah-Hartman [Fri, 22 Apr 2016 08:09:05 +0000 (17:09 +0900)]
Merge tag 'fixes-for-v4.6-rc5' of git://git./linux/kernel/git/balbi/usb into usb-linus

Felipe writes:

usb: fixes for v4.6-rc5

No more major fixes left. Out of the 6 fixes we have
here, 4 are on dwc3.

The most important is the memory leak fix in
dwc3/debugfs.c. We also have a fix for PHY handling
in suspend/resume and a fix for dwc3-omap's error
handling.

Suspend/resume also had the potential to trigger a
NULL pointer dereference on dwc3; that's also fixed
now.

Our good ol' ffs function gets a use-after-free fix
while the generic composite.c layer has a robustness
fix by making sure reserved fields of a possible SSP
device capability descriptor is cleared to 0.

8 years agox86/mm/xen: Suppress hugetlbfs in PV guests
Jan Beulich [Thu, 21 Apr 2016 06:27:04 +0000 (00:27 -0600)]
x86/mm/xen: Suppress hugetlbfs in PV guests

Huge pages are not normally available to PV guests. Not suppressing
hugetlbfs use results in an endless loop of page faults when user mode
code tries to access a hugetlbfs mapped area (since the hypervisor
denies such PTEs to be created, but error indications can't be
propagated out of xen_set_pte_at(), just like for various of its
siblings), and - once killed in an oops like this:

  kernel BUG at .../fs/hugetlbfs/inode.c:428!
  invalid opcode: 0000 [#1] SMP
  ...
  RIP: e030:[<ffffffff811c333b>]  [<ffffffff811c333b>] remove_inode_hugepages+0x25b/0x320
  ...
  Call Trace:
   [<ffffffff811c3415>] hugetlbfs_evict_inode+0x15/0x40
   [<ffffffff81167b3d>] evict+0xbd/0x1b0
   [<ffffffff8116514a>] __dentry_kill+0x19a/0x1f0
   [<ffffffff81165b0e>] dput+0x1fe/0x220
   [<ffffffff81150535>] __fput+0x155/0x200
   [<ffffffff81079fc0>] task_work_run+0x60/0xa0
   [<ffffffff81063510>] do_exit+0x160/0x400
   [<ffffffff810637eb>] do_group_exit+0x3b/0xa0
   [<ffffffff8106e8bd>] get_signal+0x1ed/0x470
   [<ffffffff8100f854>] do_signal+0x14/0x110
   [<ffffffff810030e9>] prepare_exit_to_usermode+0xe9/0xf0
   [<ffffffff814178a5>] retint_user+0x8/0x13

This is CVE-2016-3961 / XSA-174.

Reported-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Juergen Gross <JGross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Luis R. Rodriguez <mcgrof@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: stable@vger.kernel.org
Cc: xen-devel <xen-devel@lists.xenproject.org>
Link: http://lkml.kernel.org/r/57188ED802000078000E431C@prv-mh.provo.novell.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
8 years agox86/doc: Correct limits in Documentation/x86/x86_64/mm.txt
Juergen Gross [Fri, 22 Apr 2016 07:35:04 +0000 (09:35 +0200)]
x86/doc: Correct limits in Documentation/x86/x86_64/mm.txt

Correct the size of the module mapping space and the maximum available
physical memory size of current processors.

Signed-off-by: Juergen Gross <jgross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: corbet@lwn.net
Cc: linux-doc@vger.kernel.org
Link: http://lkml.kernel.org/r/1461310504-15977-1-git-send-email-jgross@suse.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
8 years agocpu/hotplug: Fix rollback during error-out in __cpu_disable()
Sebastian Andrzej Siewior [Fri, 8 Apr 2016 12:40:15 +0000 (14:40 +0200)]
cpu/hotplug: Fix rollback during error-out in __cpu_disable()

The recent introduction of the hotplug thread which invokes the callbacks on
the plugged cpu, cased the following regression:

If takedown_cpu() fails, then we run into several issues:

 1) The rollback of the target cpu states is not invoked. That leaves the smp
    threads and the hotplug thread in disabled state.

 2) notify_online() is executed due to a missing skip_onerr flag. That causes
    that both CPU_DOWN_FAILED and CPU_ONLINE notifications are invoked which
    confuses quite some notifiers.

 3) The CPU_DOWN_FAILED notification is not invoked on the target CPU. That's
    not an issue per se, but it is inconsistent and in consequence blocks the
    patches which rely on these states being invoked on the target CPU and not
    on the controlling cpu. It also does not preserve the strict call order on
    rollback which is problematic for the ongoing state machine conversion as
    well.

To fix this we add a rollback flag to the remote callback machinery and invoke
the rollback including the CPU_DOWN_FAILED notification on the remote
cpu. Further mark the notify online state with 'skip_onerr' so we don't get a
double invokation.

This workaround will go away once we moved the unplug invocation to the target
cpu itself.

[ tglx: Massaged changelog and moved the CPU_DOWN_FAILED notifiaction to the
   target cpu ]

Fixes: 4cb28ced23c4 ("cpu/hotplug: Create hotplug threads")
Reported-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: linux-s390@vger.kernel.org
Cc: rt@linutronix.de
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Anna-Maria Gleixner <anna-maria@linutronix.de>
Link: http://lkml.kernel.org/r/20160408124015.GA21960@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
8 years agoclocksource/drivers/tango-xtal: Fix boot hang due to incorrect test
Daniel Lezcano [Tue, 19 Apr 2016 13:43:02 +0000 (15:43 +0200)]
clocksource/drivers/tango-xtal: Fix boot hang due to incorrect test

Commit 0881841f7e78 introduced a regression by inverting a test check
after calling clocksource_mmio_init(). That results on the system to
hang at boot time.

Fix it by inverting the test again.

Fixes: 0881841f7e78 ("Replace code by clocksource_mmio_init")
Reported-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
8 years agoobjtool: Fix Makefile to properly see if libelf is supported
Steven Rostedt [Wed, 20 Apr 2016 15:32:35 +0000 (11:32 -0400)]
objtool: Fix Makefile to properly see if libelf is supported

When doing a make allmodconfig, I hit the following compile error:

  In file included from builtin-check.c:32:0:
  elf.h:22:18: fatal error: gelf.h: No such file or directory
  compilation terminated.
  ...

Digging into it, it appears that the $(shell ..) command in the Makefile does
not give the proper result when it fails to find -lelf, and continues to
compile objtool.

Instead, use the "try-run" makefile macro to perform the test. This gives a
proper result for both cases.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Bernd Petrovitsch <bernd@petrovitsch.priv.at>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Chris J Arges <chris.j.arges@canonical.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michal Marek <mmarek@suse.cz>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Pedro Alves <palves@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: live-patching@vger.kernel.org
Fixes: 442f04c34a1a4 ("objtool: Add tool to perform compile-time stack metadata validation")
Link: http://lkml.kernel.org/r/20160420153234.GA24032@home.goodmis.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
8 years agoMerge branch 'drm-fixes-4.6' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Fri, 22 Apr 2016 00:39:26 +0000 (10:39 +1000)]
Merge branch 'drm-fixes-4.6' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

Misc radeon and amdgpu bug fixes for 4.6.

* 'drm-fixes-4.6' of git://people.freedesktop.org/~agd5f/linux:
  amdgpu/uvd: add uvd fw version for amdgpu
  drm/amdgpu: forbid mapping of userptr bo through radeon device file
  drm/radeon: forbid mapping of userptr bo through radeon device file
  drm/amdgpu: bump the afmt limit for CZ, ST, Polaris
  drm/amdgpu: use defines for CRTCs and AMFT blocks
  drm/radeon: print a message if ATPX dGPU power control is missing
  Revert "drm/radeon: disable runtime pm on PX laptops without dGPU power control"
  drm/amdgpu/acp: fix resume on CZ systems with AZ audio
  drm/radeon: add a quirk for a XFX R9 270X
  drm/radeon: print pci revision as well as pci ids on driver load
  drm/amdgpu: when suspending, if uvd/vce was running. need to cancel delay work.
  drm/radeon: fix initial connector audio value

8 years agodrm: Loongson-3 doesn't fully support wc memory
Huacai Chen [Tue, 19 Apr 2016 11:19:11 +0000 (19:19 +0800)]
drm: Loongson-3 doesn't fully support wc memory

Signed-off-by: Huacai Chen <chenhc@lemote.com>
Cc: stable@vger.kernel.org
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoMerge branch 'linux-4.6' of git://github.com/skeggsb/linux into drm-fixes
Dave Airlie [Fri, 22 Apr 2016 00:09:33 +0000 (10:09 +1000)]
Merge branch 'linux-4.6' of git://github.com/skeggsb/linux into drm-fixes

transform feedback fix.

* 'linux-4.6' of git://github.com/skeggsb/linux:
  drm/nouveau/gr/gf100: select a stream master to fixup tfb offset queries

8 years agodrm/nouveau/gr/gf100: select a stream master to fixup tfb offset queries
Ben Skeggs [Fri, 22 Apr 2016 00:05:21 +0000 (10:05 +1000)]
drm/nouveau/gr/gf100: select a stream master to fixup tfb offset queries

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Cc: stable@vger.kernel.org
8 years agoamdgpu/uvd: add uvd fw version for amdgpu
Sonny Jiang [Mon, 18 Apr 2016 20:05:04 +0000 (16:05 -0400)]
amdgpu/uvd: add uvd fw version for amdgpu

Was previously always hardcoded to 0.

Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
8 years agodrm/amdgpu: forbid mapping of userptr bo through radeon device file
Jérôme Glisse [Tue, 19 Apr 2016 13:07:51 +0000 (09:07 -0400)]
drm/amdgpu: forbid mapping of userptr bo through radeon device file

Allowing userptr bo which are basicly a list of page from some vma
(so either anonymous page or file backed page) would lead to serious
corruption of kernel structures and counters (because we overwrite
the page->mapping field when mapping buffer).

This will already block if the buffer was populated before anyone does
try to mmap it because then TTM_PAGE_FLAG_SG would be set in in the
ttm_tt flags. But that flag is check before ttm_tt_populate in the ttm
vm fault handler.

So to be safe just add a check to verify_access() callback.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jérôme Glisse <jglisse@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
8 years agodrm/radeon: forbid mapping of userptr bo through radeon device file
Jérôme Glisse [Tue, 19 Apr 2016 13:07:50 +0000 (09:07 -0400)]
drm/radeon: forbid mapping of userptr bo through radeon device file

Allowing userptr bo which are basicly a list of page from some vma
(so either anonymous page or file backed page) would lead to serious
corruption of kernel structures and counters (because we overwrite
the page->mapping field when mapping buffer).

This will already block if the buffer was populated before anyone does
try to mmap it because then TTM_PAGE_FLAG_SG would be set in in the
ttm_tt flags. But that flag is check before ttm_tt_populate in the ttm
vm fault handler.

So to be safe just add a check to verify_access() callback.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jérôme Glisse <jglisse@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
8 years agodrm/amdgpu: bump the afmt limit for CZ, ST, Polaris
Alex Deucher [Mon, 18 Apr 2016 22:25:34 +0000 (18:25 -0400)]
drm/amdgpu: bump the afmt limit for CZ, ST, Polaris

Fixes array overflow on these chips.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
8 years agodrm/amdgpu: use defines for CRTCs and AMFT blocks
Alex Deucher [Mon, 18 Apr 2016 22:09:57 +0000 (18:09 -0400)]
drm/amdgpu: use defines for CRTCs and AMFT blocks

Prerequiste for the next patch which ups the limits.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
8 years agodrm/dp/mst: Validate port in drm_dp_payload_send_msg()
cpaul@redhat.com [Mon, 4 Apr 2016 23:58:47 +0000 (19:58 -0400)]
drm/dp/mst: Validate port in drm_dp_payload_send_msg()

With the joys of things running concurrently, there's always a chance
that the port we get passed in drm_dp_payload_send_msg() isn't actually
valid anymore. Because of this, we need to make sure we validate the
reference to the port before we use it otherwise we risk running into
various race conditions. For instance, on the Dell MST monitor I have
here for testing, hotplugging it enough times causes us to kernel panic:

[drm:intel_mst_enable_dp] 1
[drm:drm_dp_update_payload_part2] payload 0 1
[drm:intel_get_hpd_pins] hotplug event received, stat 0x00200000, dig 0x10101011, pins 0x00000020
[drm:intel_hpd_irq_handler] digital hpd port B - short
[drm:intel_dp_hpd_pulse] got hpd irq on port B - short
[drm:intel_dp_check_mst_status] got esi 00 10 00
[drm:drm_dp_update_payload_part2] payload 1 1
general protection fault: 0000 [#1] SMP
…
Call Trace:
 [<ffffffffa012b632>] drm_dp_update_payload_part2+0xc2/0x130 [drm_kms_helper]
 [<ffffffffa032ef08>] intel_mst_enable_dp+0xf8/0x180 [i915]
 [<ffffffffa0310dbd>] haswell_crtc_enable+0x3ed/0x8c0 [i915]
 [<ffffffffa030c84d>] intel_atomic_commit+0x5ad/0x1590 [i915]
 [<ffffffffa01db877>] ? drm_atomic_set_crtc_for_connector+0x57/0xe0 [drm]
 [<ffffffffa01dc4e7>] drm_atomic_commit+0x37/0x60 [drm]
 [<ffffffffa0130a3a>] drm_atomic_helper_set_config+0x7a/0xb0 [drm_kms_helper]
 [<ffffffffa01cc482>] drm_mode_set_config_internal+0x62/0x100 [drm]
 [<ffffffffa01d02ad>] drm_mode_setcrtc+0x3cd/0x4e0 [drm]
 [<ffffffffa01c18e3>] drm_ioctl+0x143/0x510 [drm]
 [<ffffffffa01cfee0>] ? drm_mode_setplane+0x1b0/0x1b0 [drm]
 [<ffffffff810f79a7>] ? hrtimer_start_range_ns+0x1b7/0x3a0
 [<ffffffff81212962>] do_vfs_ioctl+0x92/0x570
 [<ffffffff81590852>] ? __sys_recvmsg+0x42/0x80
 [<ffffffff81212eb9>] SyS_ioctl+0x79/0x90
 [<ffffffff816b4e32>] entry_SYSCALL_64_fastpath+0x1a/0xa4
RIP  [<ffffffffa012b026>] drm_dp_payload_send_msg+0x146/0x1f0 [drm_kms_helper]

Which occurs because of the hotplug event shown in the log, which ends
up causing DRM's dp helpers to drop the port we're updating the payload
on and panic.

CC: stable@vger.kernel.org
Signed-off-by: Lyude <cpaul@redhat.com>
Reviewed-by: David Airlie <airlied@linux.ie>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoMerge branch 'linux-4.6' of git://github.com/skeggsb/linux into drm-fixes
Dave Airlie [Thu, 21 Apr 2016 23:53:07 +0000 (09:53 +1000)]
Merge branch 'linux-4.6' of git://github.com/skeggsb/linux into drm-fixes

Single nouveau regression fix
* 'linux-4.6' of git://github.com/skeggsb/linux:
  drm/nouveau/kms: fix setting of default values for dithering properties

8 years agodrm/nouveau/kms: fix setting of default values for dithering properties
Ben Skeggs [Thu, 21 Apr 2016 23:09:41 +0000 (09:09 +1000)]
drm/nouveau/kms: fix setting of default values for dithering properties

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
8 years agoMerge tag 'drm-intel-fixes-2016-04-21' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Thu, 21 Apr 2016 23:09:11 +0000 (09:09 +1000)]
Merge tag 'drm-intel-fixes-2016-04-21' of git://anongit.freedesktop.org/drm-intel into drm-fixes

Hi Dave, fixes all around, all but one are cc: stable material, the most
important ones are likely the Skylake hang fixes from Mika.

* tag 'drm-intel-fixes-2016-04-21' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: Use fw_domains_put_with_fifo() on HSW
  drm/i915: Force ringbuffers to not be at offset 0
  drm/i915: Adjust size of PIPE_CONTROL used for gen8 render seqno write
  drm/i915/skl: Fix spurious gpu hang with gt3/gt4 revs
  drm/i915/skl: Fix rc6 based gpu/system hang
  drm/i915/userptr: Hold mmref whilst calling get-user-pages
  drm/i915: Fixup the free space logic in ring_prepare
  drm/i915/skl+: Use plane size for relative data rate calculation

8 years agoMerge tag 'rtc-4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux
Linus Torvalds [Thu, 21 Apr 2016 22:41:13 +0000 (15:41 -0700)]
Merge tag 'rtc-4.6-3' of git://git./linux/kernel/git/abelloni/linux

Pull RTC fixes from Alexandre Belloni:
 "A few fixes for the RTC subsystem.  The documentation fix already
  missed 4.5 so I think it is worth taking it now:

  A documentation fix for s3c and two fixes for the ds1307"

* tag 'rtc-4.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux:
  rtc: ds1307: Use irq when available for wakeup-source device
  rtc: ds1307: ds3231 temperature s16 overflow
  rtc: s3c: Document in binding that only s3c6410 needs a src clk

8 years agoMerge tag 'pm+acpi-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Thu, 21 Apr 2016 21:29:34 +0000 (14:29 -0700)]
Merge tag 'pm+acpi-4.6-rc5' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "Two fixes for issues introduced recently, one for an intel_pstate
  driver problem uncovered by the recent switch over from using timers
  and the other one for a potential cpufreq core problem related to
  system suspend/resume.

  Specifics:

   - Fix an intel_pstate driver problem causing CPUs to get stuck in the
     highest P-state when completely idle uncovered by the recent switch
     over from using timers (Rafael Wysocki).

   - Avoid attempts to get the current CPU frequency when all devices
     (like I2C controllers that may be nedded for that purpose) have
     been suspended during system suspend/resume (Rafael Wysocki)"

* tag 'pm+acpi-4.6-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: Abort cpufreq_update_current_freq() for cpufreq_suspended set
  intel_pstate: Avoid getting stuck in high P-states when idle

8 years agortc: ds1307: Use irq when available for wakeup-source device
Nishanth Menon [Tue, 19 Apr 2016 16:23:54 +0000 (11:23 -0500)]
rtc: ds1307: Use irq when available for wakeup-source device

With commit 8bc2a40730ec ("rtc: ds1307: add support for the
DT property 'wakeup-source'") we lost the ability for rtc irq
functionality for devices that are actually hooked on a real IRQ
line and have capability to wakeup as well. This is not an expected
behavior. So, instead of just not requesting IRQ, skip the IRQ
requirement only if interrupts are not defined for the device.

Fixes: 8bc2a40730ec ("rtc: ds1307: add support for the DT property 'wakeup-source'")
Reported-by: Tony Lindgren <tony@atomide.com>
Cc: Michael Lange <linuxstuff@milaw.biz>
Cc: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
8 years agortc: ds1307: ds3231 temperature s16 overflow
Zhuang Yuyao [Mon, 18 Apr 2016 00:21:42 +0000 (09:21 +0900)]
rtc: ds1307: ds3231 temperature s16 overflow

while retrieving temperature from ds3231, the result may be overflow
since s16 is too small for a multiplication with 250.

ie. if temp_buf[0] == 0x2d, the result (s16 temp) will be negative.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Tested-by: Michael Tatarinov <kukabu@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Thu, 21 Apr 2016 19:57:34 +0000 (12:57 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fix memory leak in iwlwifi, from Matti Gottlieb.

 2) Add missing registration of netfilter arp_tables into initial
    namespace, from Florian Westphal.

 3) Fix potential NULL deref in DecNET routing code.

 4) Restrict NETLINK_URELEASE to truly bound sockets only, from Dmitry
    Ivanov.

 5) Fix dst ref counting in VRF, from David Ahern.

 6) Fix TSO segmenting limits in i40e driver, from Alexander Duyck.

 7) Fix heap leak in PACKET_DIAG_MCLIST, from Mathias Krause.

 8) Ravalidate IPV6 datagram socket cached routes properly, particularly
    with UDP, from Martin KaFai Lau.

 9) Fix endian bug in RDS dp_ack_seq handling, from Qing Huang.

10) Fix stats typing in bcmgenet driver, from Eric Dumazet.

11) Openvswitch needs to orphan SKBs before ipv6 fragmentation handing,
    from Joe Stringer.

12) SPI device reference leak in spi_ks8895 PHY driver, from Mark Brown.

13) atl2 doesn't actually support scatter-gather, so don't advertise the
    feature.  From Ben Hucthings.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (72 commits)
  openvswitch: use flow protocol when recalculating ipv6 checksums
  Driver: Vmxnet3: set CHECKSUM_UNNECESSARY for IPv6 packets
  atl2: Disable unimplemented scatter/gather feature
  net/mlx4_en: Split SW RX dropped counter per RX ring
  net/mlx4_core: Don't allow to VF change global pause settings
  net/mlx4_core: Avoid repeated calls to pci enable/disable
  net/mlx4_core: Implement pci_resume callback
  net: phy: spi_ks8895: Don't leak references to SPI devices
  net: ethernet: davinci_emac: Fix platform_data overwrite
  net: ethernet: davinci_emac: Fix Unbalanced pm_runtime_enable
  qede: Fix single MTU sized packet from firmware GRO flow
  qede: Fix setting Skb network header
  qede: Fix various memory allocation error flows for fastpath
  tcp: Merge tx_flags and tskey in tcp_shifted_skb
  tcp: Merge tx_flags and tskey in tcp_collapse_retrans
  drivers: net: cpsw: fix wrong regs access in cpsw_ndo_open
  tcp: Fix SOF_TIMESTAMPING_TX_ACK when handling dup acks
  openvswitch: Orphan skbs before IPv6 defrag
  Revert "Prevent NUll pointer dereference with two PHYs on cpsw"
  VSOCK: Only check error on skb_recv_datagram when skb is NULL
  ...

8 years agoopenvswitch: use flow protocol when recalculating ipv6 checksums
Simon Horman [Thu, 21 Apr 2016 01:49:15 +0000 (11:49 +1000)]
openvswitch: use flow protocol when recalculating ipv6 checksums

When using masked actions the ipv6_proto field of an action
to set IPv6 fields may be zero rather than the prevailing protocol
which will result in skipping checksum recalculation.

This patch resolves the problem by relying on the protocol
in the flow key rather than that in the set field action.

Fixes: 83d2b9ba1abc ("net: openvswitch: Support masked set actions.")
Cc: Jarno Rajahalme <jrajahalme@nicira.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoDriver: Vmxnet3: set CHECKSUM_UNNECESSARY for IPv6 packets
Shrikrishna Khare [Thu, 21 Apr 2016 01:12:29 +0000 (18:12 -0700)]
Driver: Vmxnet3: set CHECKSUM_UNNECESSARY for IPv6 packets

For IPv6, if the device indicates that the checksum is correct, set
CHECKSUM_UNNECESSARY.

Reported-by: Subbarao Narahari <snarahari@vmware.com>
Signed-off-by: Shrikrishna Khare <skhare@vmware.com>
Signed-off-by: Jin Heo <heoj@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoatl2: Disable unimplemented scatter/gather feature
Ben Hutchings [Wed, 20 Apr 2016 22:23:08 +0000 (23:23 +0100)]
atl2: Disable unimplemented scatter/gather feature

atl2 includes NETIF_F_SG in hw_features even though it has no support
for non-linear skbs.  This bug was originally harmless since the
driver does not claim to implement checksum offload and that used to
be a requirement for SG.

Now that SG and checksum offload are independent features, if you
explicitly enable SG *and* use one of the rare protocols that can use
SG without checkusm offload, this potentially leaks sensitive
information (before you notice that it just isn't working).  Therefore
this obscure bug has been designated CVE-2016-2117.

Reported-by: Justin Yackoski <jyackoski@crypto-nite.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Fixes: ec5f06156423 ("net: Kill link between CSUM and SG features.")
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoirqchip/mips-gic: Don't overrun pcpu_masks array
Paul Burton [Thu, 21 Apr 2016 10:31:54 +0000 (11:31 +0100)]
irqchip/mips-gic: Don't overrun pcpu_masks array

Commit 2a0787051182 ("irqchip/mips-gic: Use gic_vpes instead of
NR_CPUS") & commit 78930f09b940 ("irqchip/mips-gic: Clear percpu_masks
correctly when mapping") both introduce code which accesses gic_vpes
entries in the pcpu_masks array. However, this array has length NR_CPUS.
If NR_CPUS is less than gic_vpes (ie. the kernel supports use of less
CPUs than are present in the system) then we overrun the array, clobber
some other data & generally die pretty promptly.

Most notably this affects uniprocessor kernels running on any multicore
or multithreaded Malta with a GIC (ie. the vast majority of real Malta
boards).

Fix this by only accessing up to min(gic_vpes, NR_CPUS) entries in the
pcpu_masks array, preventing the array overrun.

Fixes: 2a0787051182 ("irqchip/mips-gic: Use gic_vpes instead of NR_CPUS")
Fixes: 78930f09b940 ("irqchip/mips-gic: Clear percpu_masks correctly when mapping")
Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Link: http://lkml.kernel.org/r/1461234714-9975-1-git-send-email-paul.burton@imgtec.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
8 years agoMerge branch 'mlx4-fixes'
David S. Miller [Thu, 21 Apr 2016 19:02:41 +0000 (15:02 -0400)]
Merge branch 'mlx4-fixes'

Or Gerlitz says:

====================
Mellaox 40G driver fixes for 4.6-rc

With the fix for ARM bug being under the works, these are
few other fixes for mlx4 we have ready to go.

Eran addressed the problematic/wrong reporting of dropped packets, Daniel
fixed some matters related to PPC EEH's and Jenny's patch makes sure
VFs can't change the port's pause settings.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_en: Split SW RX dropped counter per RX ring
Eran Ben Elisha [Wed, 20 Apr 2016 13:01:18 +0000 (16:01 +0300)]
net/mlx4_en: Split SW RX dropped counter per RX ring

Count SW packet drops per RX ring instead of a global counter. This
will allow monitoring the number of rx drops per ring.

In addition, SW rx_dropped counter was overwritten by HW rx_dropped
counter, sum both of them instead to show the accurate value.

Fixes: a3333b35da16 ('net/mlx4_en: Moderate ethtool callback to [...] ')
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reported-by: Brenden Blanco <bblanco@plumgrid.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_core: Don't allow to VF change global pause settings
Eugenia Emantayev [Wed, 20 Apr 2016 13:01:17 +0000 (16:01 +0300)]
net/mlx4_core: Don't allow to VF change global pause settings

Currently changing global pause settings is done via SET_PORT
command with input modifier GENERAL. This command is allowed
for each VF since MTU setting is done via the same command.

Change the above to the following scheme: before passing the
request to the FW, the PF will check whether it was issued
by a slave. If yes, don't change global pause and warn,
otherwise change to the requested value and store for
further reference.

Signed-off-by: Eugenia Emantayev <eugenia@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_core: Avoid repeated calls to pci enable/disable
Daniel Jurgens [Wed, 20 Apr 2016 13:01:16 +0000 (16:01 +0300)]
net/mlx4_core: Avoid repeated calls to pci enable/disable

Maintain the PCI status and provide wrappers for enabling and disabling
the PCI device.  Performing the actions more than once without doing
its opposite results in warning logs.

This occurred when EEH hotplugged the device causing a warning for
disabling an already disabled device.

Fixes: 2ba5fbd62b25 ('net/mlx4_core: Handle AER flow properly')
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_core: Implement pci_resume callback
Daniel Jurgens [Wed, 20 Apr 2016 13:01:15 +0000 (16:01 +0300)]
net/mlx4_core: Implement pci_resume callback

Move resume related activities to a new pci_resume function instead of
performing them in mlx4_pci_slot_reset.  This change is needed to avoid
a hotplug during EEH recovery due to commit f2da4ccf8bd4 ("powerpc/eeh:
More relaxed hotplug criterion").

Fixes: 2ba5fbd62b25 ('net/mlx4_core: Handle AER flow properly')
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: phy: spi_ks8895: Don't leak references to SPI devices
Mark Brown [Wed, 20 Apr 2016 11:54:05 +0000 (12:54 +0100)]
net: phy: spi_ks8895: Don't leak references to SPI devices

The ks8895 driver is using spi_dev_get() apparently just to take a copy
of the SPI device used to instantiate it but never calls spi_dev_put()
to free it.  Since the device is guaranteed to exist between probe() and
remove() there should be no need for the driver to take an extra
reference to it so fix the leak by just using a straight assignment.

Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ethernet: davinci_emac: Fix platform_data overwrite
Neil Armstrong [Wed, 20 Apr 2016 08:56:45 +0000 (10:56 +0200)]
net: ethernet: davinci_emac: Fix platform_data overwrite

When the DaVinci emac driver is removed and re-probed, the actual
pdev->dev.platform_data is populated with an unwanted valid pointer saved by
the previous davinci_emac_of_get_pdata() call, causing a kernel crash when
calling priv->int_disable() in emac_int_disable().

Unable to handle kernel paging request at virtual address c8622a80
...
[<c0426fb4>] (emac_int_disable) from [<c0427700>] (emac_dev_open+0x290/0x5f8)
[<c0427700>] (emac_dev_open) from [<c04c00ec>] (__dev_open+0xb8/0x120)
[<c04c00ec>] (__dev_open) from [<c04c0370>] (__dev_change_flags+0x88/0x14c)
[<c04c0370>] (__dev_change_flags) from [<c04c044c>] (dev_change_flags+0x18/0x48)
[<c04c044c>] (dev_change_flags) from [<c052bafc>] (devinet_ioctl+0x6b4/0x7ac)
[<c052bafc>] (devinet_ioctl) from [<c04a1428>] (sock_ioctl+0x1d8/0x2c0)
[<c04a1428>] (sock_ioctl) from [<c014f054>] (do_vfs_ioctl+0x41c/0x600)
[<c014f054>] (do_vfs_ioctl) from [<c014f2a4>] (SyS_ioctl+0x6c/0x7c)
[<c014f2a4>] (SyS_ioctl) from [<c000ff60>] (ret_fast_syscall+0x0/0x1c)

Fixes: 42f59967a091 ("net: ethernet: davinci_emac: add OF support")
Cc: Brian Hutchinson <b.hutchman@gmail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ethernet: davinci_emac: Fix Unbalanced pm_runtime_enable
Neil Armstrong [Wed, 20 Apr 2016 08:56:13 +0000 (10:56 +0200)]
net: ethernet: davinci_emac: Fix Unbalanced pm_runtime_enable

In order to avoid an Unbalanced pm_runtime_enable in the DaVinci
emac driver when the device is removed and re-probed, and a
pm_runtime_disable() call in davinci_emac_remove().

Actually, using unbind/bind on a TI DM8168 SoC gives :
$ echo 4a120000.ethernet > /sys/bus/platform/drivers/davinci_emac/unbind
net eth1: DaVinci EMAC: davinci_emac_remove()
$ echo 4a120000.ethernet > /sys/bus/platform/drivers/davinci_emac/bind
davinci_emac 4a120000.ethernet: Unbalanced pm_runtime_enable

Cc: Brian Hutchinson <b.hutchman@gmail.com>
Fixes: 3ba97381343b ("net: ethernet: davinci_emac: add pm_runtime support")
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'pm-cpufreq-fixes'
Rafael J. Wysocki [Thu, 21 Apr 2016 18:57:46 +0000 (20:57 +0200)]
Merge branch 'pm-cpufreq-fixes'

* pm-cpufreq-fixes:
  cpufreq: Abort cpufreq_update_current_freq() for cpufreq_suspended set
  intel_pstate: Avoid getting stuck in high P-states when idle

8 years agoMerge branch 'qed-fixes'
David S. Miller [Thu, 21 Apr 2016 18:51:29 +0000 (14:51 -0400)]
Merge branch 'qed-fixes'

Manish Chopra says:

====================
qede: Bug fixes

This series fixes -

* various memory allocation failure flows for fastpath
* issues with respect to driver GRO packets handling

V1->V2

* Send series against net instead of net-next.

Please consider applying this series to "net"
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoqede: Fix single MTU sized packet from firmware GRO flow
Manish Chopra [Wed, 20 Apr 2016 07:03:29 +0000 (03:03 -0400)]
qede: Fix single MTU sized packet from firmware GRO flow

In firmware assisted GRO flow there could be a single MTU sized
segment arriving due to firmware aggregation timeout/last segment
in an aggregation flow, which is not expected to be an actual gro
packet. So If a skb has zero frags from the GRO flow then simply
push it in the stack as non gso skb.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Yuval Mintz <yuval.mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoqede: Fix setting Skb network header
Manish Chopra [Wed, 20 Apr 2016 07:03:28 +0000 (03:03 -0400)]
qede: Fix setting Skb network header

Skb's network header needs to be set before extracting IPv4/IPv6
headers from it.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Yuval Mintz <yuval.mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoqede: Fix various memory allocation error flows for fastpath
Manish Chopra [Wed, 20 Apr 2016 07:03:27 +0000 (03:03 -0400)]
qede: Fix various memory allocation error flows for fastpath

This patch handles memory allocation failures for fastpath
gracefully in the driver.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Yuval Mintz <yuval.mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tcp-coalesce-merge-timestamps'
David S. Miller [Thu, 21 Apr 2016 18:40:56 +0000 (14:40 -0400)]
Merge branch 'tcp-coalesce-merge-timestamps'

Martin KaFai Lau says:

====================
tcp: Merge timestamp info when coalescing skbs

This series is separated from the RFC series related to
tcp_sendmsg(MSG_EOR) and it is targeting for the net branch.
This patchset is focusing on fixing cases where TCP
timestamp could be lost after coalescing skbs.

A BPF prog is used to kprobe to sock_queue_err_skb()
and print out the value of serr->ee.ee_data.  The BPF
prog (run-able from bcc) is attached here:

BPF prog used for testing:
~~~~~

from __future__ import print_function
from bcc import BPF

bpf_text = """

int trace_err_skb(struct pt_regs *ctx)
{
struct sk_buff *skb = (struct sk_buff *)ctx->si;
struct sock *sk = (struct sock *)ctx->di;
struct sock_exterr_skb *serr;
u32 ee_data = 0;

if (!sk || !skb)
return 0;

serr = SKB_EXT_ERR(skb);
bpf_probe_read(&ee_data, sizeof(ee_data), &serr->ee.ee_data);
bpf_trace_printk("ee_data:%u\\n", ee_data);

return 0;
};
"""

b = BPF(text=bpf_text)
b.attach_kprobe(event="sock_queue_err_skb", fn_name="trace_err_skb")
print("Attached to kprobe")
b.trace_print()
====================

Signed-off-by: David S. Miller <davem@davemloft.net>