UPSTREAM: timekeeping: Add a fast and NMI safe boot clock

This boot clock can be used as a tracing clock and will account for
suspend time.

To keep it NMI safe since we're accessing from tracing, we're not using a
separate timekeeper with updates to monotonic clock and boot offset
protected with seqlocks. This has the following minor side effects:

(1) Its possible that a timestamp be taken after the boot offset is updated
but before the timekeeper is updated. If this happens, the new boot offset
is added to the old timekeeping making the clock appear to update slightly
earlier:
   CPU 0                                        CPU 1
   timekeeping_inject_sleeptime64()
   __timekeeping_inject_sleeptime(tk, delta);
                                                timestamp();
   timekeeping_update(tk, TK_CLEAR_NTP...);

(2) On 32-bit systems, the 64-bit boot offset (tk->offs_boot) may be
partially updated.  Since the tk->offs_boot update is a rare event, this
should be a rare occurrence which postprocessing should be able to handle.

Bug: b/33184060

Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Joel Fernandes <joelaf@google.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>

ANDROID: goldfish_pipe: fix allmodconfig build

tree:   https://android.googlesource.com/kernel/common android-4.4
head:   6297c6ba0d217d5b0998738fbfaff2f04cad77e6
commit: bc43565e1ac5ba3f204886a2275726bb4c3d44e6 [18/20] ANDROID:
goldfish_pipe: An implementation of more parallel pipe
config: i386-randconfig-s1-201646 (attached as .config)
compiler: gcc-6 (Debian 6.2.0-3) 6.2.0 20160901
reproduce:
        git checkout bc43565e1ac5ba3f204886a2275726bb4c3d44e6
        # save the attached .config to linux build tree
        make ARCH=i386

All errors (new ones prefixed by >>):

>> ERROR: "goldfish_pipe_device_deinit_v1" [drivers/platform/goldfish/goldfish_pipe_v2.ko] undefined!
>> ERROR: "goldfish_pipe_device_init_v1" [drivers/platform/goldfish/goldfish_pipe_v2.ko] undefined!
>> ERROR: "pipe_dev" [drivers/platform/goldfish/goldfish_pipe.ko] undefined!

Change-Id: Ibd51441edf82e6bb6824acc05ea795570cc374e8

ANDROID: goldfish: goldfish_pipe: fix locking errors

If the get_user_pages_fast() call in goldfish_pipe_read_write() failed,
it would return while still holding pipe->lock.

goldfish_pipe_read_write() later releases and tries to re-acquire
pipe->lock.  If the re-acquire call failed, goldfish_pipe_read_write()
would try unlock pipe->lock on exit anyway.

This fixes the smatch messages:

drivers/platform/goldfish/goldfish_pipe.c:392 goldfish_pipe_read_write() error: double unlock 'mutex:&pipe->lock'
drivers/platform/goldfish/goldfish_pipe.c:397 goldfish_pipe_read_write() warn: inconsistent returns 'mutex:&pipe->lock'.

Change-Id: Ifd06a76b32027ca451a001704ade0c5440ed69c4
Signed-off-by: Greg Hackmann <ghackmann@google.com>

ANDROID: video: goldfishfb: fix platform_no_drv_owner.cocci warnings

drivers/video/fbdev/goldfishfb.c:318:3-8: No need to set .owner here. The core will do it.

 Remove .owner field if calls are used which set it automatically

Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci

CC: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Guenter Roeck <groeck@chromium.org>

ANDROID: goldfish_pipe: fix call_kern.cocci warnings

Function get_free_pipe_id_locked called on line 671 inside lock on line
669 but uses GFP_KERNEL.  Replace with GFP_ATOMIC.

Generated by: scripts/coccinelle/locks/call_kern.cocci

CC: Yurii Zubrytskyi <zyy@google.com>
Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Guenter Roeck <groeck@chromium.org>

arm64: rename ranchu defconfig to ranchu64

Change-Id: Ib7cd1ef722167905957623f65c3cc064e9d5c357

ANDROID: arch: x86: disable pic for Android toolchain

Android toolchains enable PIC, so explicitly disable it with
-fno-pic (this is the upstream gcc default)

Signed-off-by: Greg Hackmann <ghackmann@google.com>
(cherry picked from commit 892606ece2bebfa5a1ed62e9552cc973707ae9d3)

Change-Id: I1e600363e5d18e459479fe4eb23d76855e16868d

ANDROID: goldfish_pipe: An implementation of more parallel pipe

This is a driver code for a redesigned android pipe.
Currently it works for x86 and x64 emulators with the following
performance results:
  ADB push to /dev/null,
  Ubuntu,
  400 MB file,
  times are for 1/10/100 parallel adb commands
x86 adb push: (4.4s / 11.5s / 2m10s) -> (2.8s / 6s / 51s)
x64 adb push: (7s / 15s / (too long, 6m+) -> (2.7s / 6.2s / 52s)

ADB pull and push to /data/ have the same %% of speedup
More importantly, I don't see any signs of slowdowns when
run in parallel with Antutu benchmark, so it is definitely
making much better job at multithreading.

The code features dynamic host detection: old emulator gets
the previous version of the pipe driver code.

Combine follow patch from android-goldfish-3.10

b543285 [pipe] Increase the default pipe buffers size, make it configurable

Signed-off-by: "Yurii Zubrytskyi" <zyy@google.com>
Change-Id: I140d506204cab6e78dd503e5a43abc8886e4ffff

ANDROID: goldfish_pipe: bugfixes and performance improvements.

Combine following patches from android-goldfish-3.18 branch:

c0f015a [pipe] Fix the pipe driver for x64 platform + correct pages count
48e6bf5 [pipe] Use get_use_pages_fast() which is possibly faster
fb20f13 [goldfish] More pages in goldfish pipe
f180e6d goldfish_pipe: Return from read_write on signal and EIO
3dec3b7 [pipe] Fix a minor leak in setup_access_params_addr()

Change-Id: I1041fd65d7faaec123e6cedd3dbbc5a2fbb86c4d

ANDROID: goldfish: Add goldfish sync driver

This is kernel driver for controlling the Goldfish sync
device on the host. It is used to maintain ordering
in critical OpenGL state changes while using
GPU emulation.

The guest open()'s the Goldfish sync device to create
a context for possibly maintaining sync timeline and fences.
There is a 1:1 correspondence between such sync contexts
and OpenGL contexts in the guest that need synchronization
(which in turn, is anything involving swapping buffers,
SurfaceFlinger, or Hardware Composer).

The ioctl QUEUE_WORK takes a handle to a sync object
and attempts to tell the host GPU to wait on the sync object
and deal with signaling it. It possibly outputs
a fence FD on which the Android systems that use them
(GLConsumer, SurfaceFlinger, anything employing
EGL_ANDROID_native_fence_sync) can use to wait.

Design decisions and work log:

- New approach is to have the guest issue ioctls that
  trigger host wait, and then host increments timeline.
- We need the host's sync object handle and sync thread handle
  as the necessary information for that.
- ioctl() from guest can work simultaneously with the
  interrupt handling for commands from host.
- optimization: don't write back on timeline inc
- Change spin lock design to be much more lightweight;
  do not call sw_sync functions or loop too long
  anywhere.
- Send read/write commands in batches to minimize guest/host
  transitions.
- robustness: BUG if we will overrun the cmd buffer.
- robustness: return fd -1 if we cannot get an unused fd.
- correctness: remove global mutex
- cleanup pass done, incl. but not limited to:
    - removal of clear_upto and
    - switching to devm_***

This is part of a sequential, multi-CL change:

external/qemu:

https://android-review.googlesource.com/239442 <- host-side device's
host interface

https://android-review.googlesource.com/221593
https://android-review.googlesource.com/248563
https://android-review.googlesource.com/248564
https://android-review.googlesource.com/223032

external/qemu-android:

https://android-review.googlesource.com/238790 <- host-side device
implementation

kernel/goldfish:

https://android-review.googlesource.com/232631 <- needed
https://android-review.googlesource.com/238399 <- this CL

Also squash following bug fixes from android-goldfish-3.18 branch.

b44d486 goldfish_sync: provide a signal to detect reboot
ad1f597 goldfish_sync: fix stalls by avoiding early kfree()
de208e8 [goldfish-sync] Fix possible race between kernel and user space

Change-Id: I22f8a0e824717a7e751b1b0e1b461455501502b6

ANDROID: goldfish: add ranchu defconfigs

Change-Id: I73ef1b132b6203ae921a1e1d4850eaadf58f8926

ANDROID: goldfish_audio: Clear audio read buffer status after each read

The buffer_status field is interrupt updated. After every read request,
the buffer_status read field should be reset so that on the next loop
iteration we don't read a stale value and read data before the
device is ready.

Signed-off-by: “Joshua Lang” <joshualang@google.com>
Change-Id: I4943d5aaada1cad9c7e59a94a87c387578dabe86

ANDROID: goldfish_events: no extra EV_SYN; register goldfish

If we send SYN_REPORT on every single
multitouch event, it breaks the multitouch.

The multitouch becomes janky and
having to click 2-3 times to
do stuff (plus randomly activating notification
bars when not clicking)

If we suppress these SYN_REPORTS,
multitouch will work fine, plus the events
will have a protocol that looks nice.

In addition, we need to register Goldfish Events
as a multitouch device by issuing
input_mt_init_slots, otherwise
input_handle_abs_event in drivers/input/input.c
will silently drop all ABS_MT_SLOT events,
making it so that touches with more than 1 finger
do not work properly.

Signed-off-by: "Lingfeng Yang" <lfy@google.com>
Change-Id: Ib2350f7d1732449d246f6f0d9b7b08f02cc7c2dd
(cherry picked from commit 6cf40d0a16330e1ef42bdf07d9aba6c16ee11fbc)

ANDROID: goldfish_fb: Set pixclock = 0

User space Android code identifies pixclock == 0 as a sign for emulation
and will set the frame rate to 60 fps when reading this value, which is
the desired outcome.

Change-Id: I759bf518bf6683446bc786bf1be3cafa02dd8d42
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

ANDROID: goldfish: Enable ACPI-based enumeration for goldfish audio

Follow the same way in which ACPI was enabled for goldfish battery. See
commit d3be10e for details.

Change-Id: I6ffe38ebc80fb8af8322152370b9d1fd227eaf50
Signed-off-by: Yu Ning <yu.ning@intel.com>

ANDROID: goldfish: Enable ACPI-based enumeration for goldfish framebuffer

Follow the same way in which ACPI was enabled for goldfish battery. See
commit d3be10e for details.

Note that this patch also depends on commit af33cac.

Change-Id: Ic63b6e7e0a4b9896ef9a9d0ed135a7796a4c1fdb
Signed-off-by: Yu Ning <yu.ning@intel.com>

ANDROID: video: goldfishfb: add devicetree bindings

Change-Id: I5f4ba861b981edf39af537001f8ac72202927031
Signed-off-by: Greg Hackmann <ghackmann@google.com>

BACKPORT: staging: goldfish: audio: fix compiliation on arm

We do actually need slab.h, by luck we get it on other platforms but not
always on ARM. Include it properly.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 4532150762ceb0d6fd765ebcb3ba6966fbb8faab)

Change-Id: I93a0c35da40f26aaa7c253e3c0cefaa883ea3391

BACKPORT: Input: goldfish_events - enable ACPI-based enumeration for goldfish events

Add ACPI binding to the goldfish events driver.

Signed-off-by: Jason Hu <jia-cheng.hu@intel.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan <alan@linux.intel.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
(cherry picked from commit 0581ce09fd2c976125a20791268d7206db156d2f)

Change-Id: Ic3e4f1cffb111ea6c69977e63dd598e3fcb55f19

BACKPORT: goldfish: Enable ACPI-based enumeration for goldfish battery

Add the ACPI bindings to the goldfish battery driver.

Signed-off-by: Yu Ning <yu.ning@intel.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Sebastian Reichel <sre@kernel.org>
(cherry picked from commit fdb2f37a54470473c6b7c9d680c4c114dd9bc434)

Change-Id: I3b53481b5868b0b26848397420c9ba16a747819f

BACKPORT: drivers: tty: goldfish: Add device tree bindings

Enable support for registering this device using the device tree.
Device tree node example for registering Goldfish TTY device :

goldfish_tty@1f004000 {
    interrupts = <0xc>;
    reg = <0x1f004000 0x1000>;
    compatible = "google,goldfish-tty";
};

Signed-off-by: Miodrag Dinic <miodrag.dinic@imgtec.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 9b883eea26ccf043b608e398cf6a26231d44f5fb)

Change-Id: Idbe1bbac4f371e2feb6730712b08b66be1188ea7

BACKPORT: tty: goldfish: support platform_device with id -1

When the platform bus sets the platform_device id to -1 (PLATFORM_DEVID_NONE),
use an incrementing counter for the TTY index instead

Signed-off-by: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 465893e18878e119d8d0255439fad8debbd646fd)

Change-Id: Ifec5ee9d71c7c076e59bb7af77c0184d1b1383cb

BACKPORT: Input: goldfish_events - add devicetree bindings

Add device tree bindings to the Goldfish virtual platform event driver.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan <alan@linux.intel.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
(cherry picked from commit 8c5dc5a1ada2b79259e55a4bd150135d23529c6a)

Change-Id: I677d8e0d92294f53f7cc5a79300b6462b65e8aad

BACKPORT: power: goldfish_battery: add devicetree bindings

Add device tree bindings to the Goldfish virtual platform battery drivers.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Sebastian Reichel <sre@kernel.org>
(cherry picked from commit 65d687a7b7d6f27e4306fe8cc8a1ca66a1a760f6)

Change-Id: If947ea3341ff0cb713c56e14d18d51a3f5912b64

BACKPORT: staging: goldfish: audio: add devicetree bindings

Introduce devicetree bindings to the Goldfish staging audio driver.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
Signed-off-by: Jin Qian <jinqian@android.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 283ded10312a3b75e384313f6f529ec2c636cf2c)

Change-Id: Ib75d3a4cac7353084a8da18a96fb298a759bacc0

ANDROID: usb: gadget: function: cleanup: Add blank line after declaration

Fix warning generated by checkpatch.pl:
Missing a blank line after declarations

Change-Id: Id129bb8cc8fa37c67a647e2e5996bb2817020e65
Signed-off-by: Anson Jacob <ansonjacob.aj@gmail.com>

cpufreq: sched: Fix kernel crash on accessing sysfs file

If the cpufreq driver hasn't set the CPUFREQ_HAVE_GOVERNOR_PER_POLICY
flag, then the kernel will crash on accessing sysfs files for the sched
governor.

CPUFreq governors we can have the governor specific sysfs files in two
places:

A. /sys/devices/system/cpu/cpuX/cpufreq/<governor>
B. /sys/devices/system/cpu/cpufreq/<governor>

The case A. is for governor per policy case, where we can control the
governor tunables for each policy separately. The case B. is for system
wide tunable values.

The schedfreq governor only implements the case A. and not B.  The sysfs
files in case B will still be present in
/sys/devices/system/cpu/cpufreq/<governor>, but accessing them will
crash kernel as the governor doesn't support that.

Moreover the sched governor is pretty new and will be used only for the
ARM platforms and there is no need to support the case B at all.

Hence use policy->kobj instead of get_governor_parent_kobj(), so that we
always create the sysfs files in path A.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>

usb: gadget: f_mtp: simplify ptp NULL pointer check

Simplify MTP/PTP dev NULL pointer check introduced in
Change-Id: Ic44a699d96df2e13467fc081bff88b97dcc5afb2
and restrict it to MTP/PTP function level only.

Return ERR_PTR() instead of NULL from mtp_ptp function
to skip doing NULL pointer checks all the way up to
configfs.c

Fixes: Change-Id: Ic44a699d96df2e13467fc081bff88b97dcc5afb2
       ("usb: gadget: fix NULL ptr derefer while symlinking PTP func")
Change-Id: Iab7c55089c115550c3506f6cca960a07ae52713d
Signed-off-by: Amit Pundir <amit.pundir@linaro.org>

drm/rockchip: boot_logo: encoder atomic_check before enable crtc

We need update encoder output_mode and output_type before enable crtc.

Change-Id: If0c05b4d254dcac2abd6fc024fc3ffc1c5f02323
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>

camera: rockchip: camsys_drv v0.0x21.7

Change-Id: Id8d254f74b3eeb251d7369dc866f32f026d606df
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

arm64: dts: rockchip: rk3399-android-next: add isp

Change-Id: I8db91aae4dfd9c1b1ad39adb1e8377aba0c34fca
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rockchip: rk3399: add iommu for isp

Change-Id: Ic405ab5877355ed4128e3f473c21acdf5d026d1d
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM: dts: rockchip: enable the vpu service for rk3288 evb

Both VPU and HEVC service are enabled.

Change-Id: Id6dc1f0139636e8b2016b9194adddc3f87ec4e19
Signed-off-by: Randy Li <randy.li@rock-chips.com>

ARM: dts: rockchip: add hevc vpu service for rk3288

Change-Id: I87a8c9df636e04b92948c87c27e82b43a67de184
Signed-off-by: Randy Li <randy.li@rock-chips.com>

camera: camsys_drv: v0.0x21.6 support drm

Change-Id: If0305388c5b445adbcf693504849a6be000e64a9
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rk3399-evb-rev1-android: enable isp0&isp1

Change-Id: I88dda6f8e90f549ff1bbe791acec94eae30fdd45
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

UPSTREAM: iommu/rockchip: Don't feed NULL res pointers to devres

If we do, devres prints a "invalid resource" string in the error
loglevel.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit 8d7f2d84ed2d44b05e1ce88fa4b74886af46a139)

Change-Id: I3849bcaa4ceaea3247a8169b2a123a834011fbc5
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

Revert "iommu/rockchip: fix bool operation error and probe warning"

This reverts commit e0492807e6b91becccbcab6eca1df27da55d726b.

Same issue fixed by upstream(e04928 "iommu/rockchip: fix bool operation
error and probe warning) which picked as (2e7026 "UPSTREAM: iommu/rockchip:
Fix "is stall active" check).

Conflicts:
drivers/iommu/rockchip-iommu.c

Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Change-Id: I56e61894b1da14bce78ae1b3d08158dfb5b027bb

CHROMIUM: iommu/rockchip: Fix TLB flush of secondary IOMMUs

Due to the bug in current code, only first IOMMU has the TLB lines
flushed in rk_iommu_zap_lines. This patch fixes the inner loop to
execute for all IOMMUs and properly flush the TLB.

BUG=chrome-os-partner:55135
TEST=compile

Change-Id: Ica2d4b0cc3d3cbc88c70ad541dc00883f1b4e90c
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/371098
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Randy Li <randy.li@rock-chips.com>

drm/rockchip: vop: fix scale abnormal after window power on/off

Change-Id: Ifcaddf2f2b1c9c031bdf28ceb80468cfb79ce52b
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>

camera: rockchip: porting ov8858&ov4689&ov2710 driver

Change-Id: Ia9a0a39d347a9fd506f493c6639785267233e0ac
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rk3399: move opp tables to rk3399-opp.dtsi

Add a new dtsi file - rk3399-opp.dtsi, to configure opp-tables
for cpu, gpu and dmc.

Add rk3399-early-opp.dtsi for board with ES1, which need limit
frequency for cpu, gpu and dmc.

Change-Id: Ib57761fd5f405b0e79039d7a01e6e023d6f5dc2c
Reviewed-by: Finley Xiao <finley.xiao@rock-chips.com>
Reviewed-by: Huang, Tao <huangtao@rock-chips.com>
Signed-off-by: Jianqun Xu <jay.xu@rock-chips.com>

ARM: dts: rockchip: add vpu service for RK3288

It is not harm to keep both vpu service and vpu v4l2 at the same time,
that the board file choose a driver implementation and interace to
be used.

Change-Id: If79deac5bf19395cfdca821f20486985e3034389
Signed-off-by: Randy Li <randy.li@rock-chips.com>

ARM: dts: enable SD and GMAC at rk3288-evb

There is a bug in GMAC IP, it supports RGMII clock but not a speed
mode for it.

Change-Id: I8e5cca355c30920db37400901d3411eebca711ae
Signed-off-by: Randy Li <randy.li@rock-chips.com>

isp10: rockchip: v0.1.6

Change-Id: I1b702fb51d1baabd47f190fdafd79ecd22e18be0
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rk3399-evb-rev3-android: enable isp0&isp1

Change-Id: Ib2d4f304e2fc4589d55e60d452571ed46e5dae4d
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rk3399-evb-rev2-android: enable isp0&isp1

Change-Id: Ibd5244985eb707ca9119191cf019b56ae9055eb4
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

ARM64: dts: rk3399-android: add isp0 node for isp10

Change-Id: I1324bd021ec87f10ad4b5fd200bdf83efd1dab66
Signed-off-by: dalong.zhang <dalon.zhang@rock-chips.com>

Merge branch 'linux-linaro-lsk-v4.4' into linux-linaro-lsk-v4.4-android

Merge remote-tracking branch 'origin/v4.4/topic/wb-cg2' into linux-linaro-lsk-v4.4

cgroup: replace unified-hierarchy.txt with a proper cgroup v2 documentation

Now that cgroup v2 is almost out of the door, replace the development
documentation unified-hierarchy.txt with Documentation/cgroup.txt
which is a superset of unified-hierarchy.txt and authoritatively
describes all userland-visible aspects of cgroup.

v2: Updated to include all information from blkio-controller.txt and
    list filesystems which support cgroup writeback as suggested by
    Vivek.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
(cherry picked from commit 6c2920926b10e8303378408e3c2b8952071d4344)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

cgroup: rename Documentation/cgroups/ to Documentation/cgroup-legacy/

In preparation for adding cgroup2 documentation, rename
Documentation/cgroups/ to Documentation/cgroup-legacy/.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
(cherry picked from commit 0d942766453f3d23a51e0a2d430340a178b0903e)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

cgroup: replace __DEVEL__sane_behavior with cgroup2 fs type

With major controllers - cpu, memory and io - shaping up for the
unified hierarchy, cgroup2 is about ready to be, gradually, released
into the wild.  Replace __DEVEL__sane_behavior flag which was used to
select the unified hierarchy with a separate filesystem type "cgroup2"
so that unified hierarchy can be mounted as follows.

  mount -t cgroup2 none $MOUNT_POINT

The cgroup2 fs has its own magic number - 0x63677270 ("cgrp").

v2: Assign a different magic number to cgroup2 fs.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
(cherry picked from commit 67e9c74b8a873408c27ac9a8e4c1d1c8d72c93ff)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

writeback: initialize inode members that track writeback history

inode struct members that track cgroup writeback information
should be reinitialized when inode gets allocated from
kmem_cache. Otherwise, their values remain and get used by the
new inode.

Signed-off-by: Tahsin Erdogan <tahsin@google.com>
Acked-by: Tejun Heo <tj@kernel.org>
Fixes: d10c80955265 ("writeback: implement foreign cgroup inode bdi_writeback switching")
Signed-off-by: Jens Axboe <axboe@fb.com>
(cherry picked from commit 3d65ae4634ed8350aee98a4e6f4e41fe40c7d282)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

mm: page_alloc: generalize the dirty balance reserve

The dirty balance reserve that dirty throttling has to consider is
merely memory not available to userspace allocations.  There is nothing
writeback-specific about it.  Generalize the name so that it's reusable
outside of that context.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit a8d0143730d7b42c9fe6d1435d92ecce6863a62a)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

block: fix module reference leak on put_disk() call for cgroups throttle

get_disk(),get_gendisk() calls have non explicit side effect: they
increase the reference on the disk owner module.

The following is the correct sequence how to get a disk reference and
to put it:

    disk = get_gendisk(...);

    /* use disk */

    owner = disk->fops->owner;
    put_disk(disk);
    module_put(owner);

fs/block_dev.c is aware of this required module_put() call, but f.e.
blkg_conf_finish(), which is located in block/blk-cgroup.c, does not put
a module reference.  To see a leakage in action cgroups throttle config
can be used.  In the following script I'm removing throttle for /dev/ram0
(actually this is NOP, because throttle was never set for this device):

    # lsmod | grep brd
    brd                     5175  0
    # i=100; while [ $i -gt 0 ]; do echo "1:0 0" > \
        /sys/fs/cgroup/blkio/blkio.throttle.read_bps_device; i=$(($i - 1)); \
    done
    # lsmod | grep brd
    brd                     5175  100

Now brd module has 100 references.

The issue is fixed by calling module_put() just right away put_disk().

Signed-off-by: Roman Pen <roman.penyaev@profitbricks.com>
Cc: Gi-Oh Kim <gi-oh.kim@profitbricks.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: linux-block@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@fb.com>
(cherry picked from commit 39a169b62b415390398291080dafe63aec751e0a)
Signed-off-by: Alex Shi <alex.shi@linaro.org>

devfreq: rockchip: dmc: support sharing regulator with other devices

If the regulator is shared between several devices then the lowest
request voltage that meets the system constraints will be used.

Change-Id: I9645f3f81004c7203769a92367513d9d177504b2
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

MALI: midgard: support sharing regulator with other devices

If the regulator is shared between several devices then the lowest
request voltage that meets the system constraints will be used.

Change-Id: I7dda43b24c7e19098db65b51ae0c4386b46ee0b7
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

devfreq: rockchip: avoid DDR voltage domain keeping the initial voltage

If there is only one opp whose frequency is equal to the initial value
in opp table list, the DDR voltage domain will keep the initial voltage,
it may be too large.

Change-Id: I9e75d54bdc7d909baa72667821ff30beb4d62e27
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

MALI: midgard: avoid GPU voltage domain keeping the initial voltage

If there is only one opp whose frequency is equal to the initial value
in opp table list, the GPU voltage domain will keep the initial voltage,
it may be too large.

Change-Id: I87a5fb82eaac8466123b61e39a5d7587da3066da
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

PM / AVS: rockchip-cpu-avs: support adjusting initial frequency and voltage

Change-Id: I377b7fccb90ecf350a37e4609bdc8f51c4e15e7a
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

cpufreq: dt: delete flag CPUFREQ_NEED_INITIAL_FREQ_CHECK

As there are still some limitations, we prefer to implement it ourselves.

Change-Id: Ic801ed0a137b025296144cb3d8e47bcb0f8c0567
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

PM / OPP: Add dev_pm_opp_check_initial_rate()

Bootloader or kernel sets CPU frequency to an initial value before cpufreq
starts on rockchip platform, if cpu's opp table is modified to a specified
value, it will cause an issue.

For example, the initial frequency is 816MHz and voltage set by hardware
is 900mV:
1. there is only one opp whose frequency is 816MHz and voltage is 850mV
in opp table list, as they frequency is equal, the voltage will not be
changed, it is still 900mV and a little too large relative to 850mV.
2. there is only one opp whose frequency is 1200MHz and voltage is 1100mV
in opp table list, as it doesn't set voltage to 1100mV before set frequency
to 1200MHz in the dev_pm_opp_set_rate function, the initial voltage 900mV
cann't supply for 1200MHz, the system crash.

Change-Id: Id8c5efc34d9c94ff37921b33f5a76e059240d368
Signed-off-by: Finley Xiao <finley.xiao@rock-chips.com>

UPSTREAM: iommu/rockchip: Prepare to support generic DMA mapping

Set geometry for allocated domains and fix .domain_alloc() callback to
work with IOMMU_DOMAIN_DMA domain type, which is used for implicit
domains on ARM64.

Signed-off-by: Shunqian Zheng <zhengsq@rock-chips.com>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit a93db2f22b6b48369acb72f66a0ae47ec17a0b05)

Change-Id: Ib04827afadbfb32ca52c6842cd056952269cbe93
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

UPSTREAM: iommu/rockchip: Use DMA API to manage coherency

Use DMA API instead of architecture internal functions like
__cpuc_flush_dcache_area() etc.

The biggest difficulty here is that dma_map and _sync calls require some
struct device, while there is no real 1:1 relation between an IOMMU
domain and some device. To overcome this, a simple platform device is
registered for each allocated IOMMU domain.

With this patch, this driver can be used on both ARM and ARM64
platforms, such as RK3288 and RK3399 respectively.

Signed-off-by: Shunqian Zheng <zhengsq@rock-chips.com>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit 4f0aba676735c653b4e739b760c1e66cd520d3e3)

Conflicts:
drivers/iommu/rockchip-iommu.c

Change-Id: I0424318ed0cea947e7c8f8d3b52f716f6cc98ce0
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

UPSTREAM: iommu/rockchip: Fix allocation of bases array in driver probe

In .probe(), devm_kzalloc() is called with size == 0 and works only
by luck, due to internal behavior of the allocator and the fact
that the proper allocation size is small. Let's use proper value for
calculating the size.

Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi slaves")
Signed-off-by: Shunqian Zheng <zhengsq@rock-chips.com>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit 3d08f434bd58656ae630376d0b5afd6ca1ffb013)

Change-Id: I78db8fbf3cb781745a05f8bee492dd7e8ac784c5
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

UPSTREAM: iommu/rockchip: Fix zap cache during device attach

rk_iommu_command() takes a struct rk_iommu and iterates over the slave
MMUs, so this is doubly wrong in that we're passing in the wrong pointer
and talking to MMUs that we shouldn't be.

Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi slaves")
Cc: stable@vger.kernel.org
Signed-off-by: John Keeping <john@metanate.com>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit ae8a7910fb0568531033bd6ebe65590f7a4fa6e2)

Change-Id: I5d6f5dd49ad0f79facee8d345c5058af80226f83
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

UPSTREAM: iommu/rockchip: Fix "is stall active" check

Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi
slaves") rk_iommu_is_stall_active() always returns false because the
bitwise AND operates on the boolean flag promoted to an integer and a
value that is either zero or BIT(2).

Explicitly convert the right-hand value to a boolean so that both sides
are guaranteed to be either zero or one.

rk_iommu_is_paging_enabled() does not suffer from the same problem since
RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same change
for consistency and to make it clear that it's correct without needing
to lookup the value.

Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi slaves")
Signed-off-by: John Keeping <john@metanate.com>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit fbedd9b9905c1643b9f7244d88999e39632bbd87)

Conflicts:
drivers/iommu/rockchip-iommu.c

Change-Id: I5a43eb19d515eba7daf1dc4b1592ac692c115df0
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>

UPSTREAM: arm64: dts: rockchip: add gmac needed pclk for rk3399 pd

This patch fixes that sometimes hang at start-up time of the system.
As the below log:
...
[   11.136543] calling  pm_genpd_debug_init+0x0/0x60 @ 1
[   11.141602] initcall pm_genpd_debug_init+0x0/0x60 returned 0 after 11 usecs
[   11.148558] calling  genpd_poweroff_unused+0x0/0x84 @ 1
<hang>

In some cases, the rk3399 should turn off the gmac power domain to save
power if some boards didn't register the gmac device node for rk3399.
Then, rk3399 need to make sure the gmac's pclk enabled if we need
operate the gmac power domain. (Due to the NOC had enabled always)

Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(am from git/mmind/linux-rockchip.git branch for-next
 commit 2afc1db0c5bad5da75556889ebc3e75661be9028)

Change-Id: I8425b83e617de9eafaa093c3342b8e6082eb4112
Signed-off-by: Jianqun Xu <jay.xu@rock-chips.com>

Merge branch 'linux-linaro-lsk-v4.4' into linux-linaro-lsk-v4.4-android

 Merge tag 'v4.4.35' into linux-linaro-lsk-v4.4

 This is the 4.4.35 stable release

drm/rockchip: vop: fix display flash when switch to iommu mapping

Do iommu mapping with looping vop vblank register have a problem, when
iommu attach take too long time or vblank time is too short, the display
would flash.

This patch use another method to fix this problem:
  Use standby and dsp_hold interrupt to enter vblank time, when iommu
  mapping is finish, exit vop standby, then vop would start scanout
  immediately,
      vop enter standby    -> |
      dsp_hold irq         -> |--------- vblank start
      do iommu mapping     -> | vblank time
      exit standby         -> |--------- vblank end

We try add 20ms delay to iommu attach, display also looks good, that
means this method have higher compatibility.

Change-Id: I59d57c9085631d0c42174ea18890c80e26b42d22
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>

drm: rockchip: sync rga driver from 3.14

Change-Id: I503006eea09a9352186eeac645f03f513213c148
Signed-off-by: Jacob Chen <jacob2.chen@rock-chips.com>

Linux 4.4.35

netfilter: nft_dynset: fix element timeout for HZ != 1000

commit a8b1e36d0d1d6f51490e7adce35367ed6adb10e7 upstream.

With HZ=100 element timeout in dynamic sets (i.e. flow tables) is 10 times
higher than configured.

Add proper conversion to/from jiffies, when interacting with userspace.

I tested this on Linux 4.8.1, and it applies cleanly to current nf and
nf-next trees.

Fixes: 22fe54d5fefc ("netfilter: nf_tables: add support for dynamic set updates")
Signed-off-by: Anders K. Pedersen <akp@cohaesio.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/cm: Mark stale CM id's whenever the mad agent was unregistered

commit 9db0ff53cb9b43ed75bacd42a89c1a0ab048b2b0 upstream.

When there is a CM id object that has port assigned to it, it means that
the cm-id asked for the specific port that it should go by it, but if
that port was removed (hot-unplug event) the cm-id was not updated.
In order to fix that the port keeps a list of all the cm-id's that are
planning to go by it, whenever the port is removed it marks all of them
as invalid.

This commit fixes a kernel panic which happens when running traffic between
guests and we force reboot a guest mid traffic, it triggers a kernel panic:

 Call Trace:
  [<ffffffff815271fa>] ? panic+0xa7/0x16f
  [<ffffffff8152b534>] ? oops_end+0xe4/0x100
  [<ffffffff8104a00b>] ? no_context+0xfb/0x260
  [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30
  [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0
  [<ffffffff81084240>] ? process_timeout+0x0/0x10
  [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20
  [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480
  [<ffffffff81065df0>] ? default_wake_function+0x0/0x20
  [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core]
  [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core]
  [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0
  [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0
  [<ffffffff8152a815>] ? page_fault+0x25/0x30
  [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm]
  [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm]
  [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm]
  [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm]
  [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib]
  [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib]
  [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib]
  [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0
  [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40
  [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0
  [<ffffffff8109aef6>] ? kthread+0x96/0xa0
  [<ffffffff8100c20a>] ? child_rip+0xa/0x20
  [<ffffffff8109ae60>] ? kthread+0x0/0xa0
  [<ffffffff8100c200>] ? child_rip+0x0/0x20

Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation")
Signed-off-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/uverbs: Fix leak of XRC target QPs

commit 5b810a242c28e1d8d64d718cebe75b79d86a0b2d upstream.

The real QP is destroyed in case of the ref count reaches zero, but
for XRC target QPs this call was missed and caused to QP leaks.

Let's call to destroy for all flows.

Fixes: 0e0ec7e0638e ('RDMA/core: Export ib_open_qp() to share XRC...')
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/core: Avoid unsigned int overflow in sg_alloc_table

commit 3c7ba5760ab8eedec01159b267bb9bfcffe522ac upstream.

sg_alloc_table gets unsigned int as parameter while the driver
returns it as size_t. Check npages isn't greater than maximum
unsigned int.

Fixes: eeb8461e36c9 ("IB: Refactor umem to use linear SG table")
Signed-off-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/mlx5: Fix fatal error dispatching

commit dbaaff2a2caa03d472b5cc53a3fbfd415c97dc26 upstream.

When an internal error condition is detected, make sure to set the
device inactive after dispatching the event so ULPs can get a
notification of this event.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Reviewed-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/mlx5: Use cache line size to select CQE stride

commit 16b0e0695a73b68d8ca40288c8f9614ef208917b upstream.

When creating kernel CQs use 128B CQE stride if the
cache line size is 128B, 64B otherwise.  This prevents
multiple CQEs from residing in a 128B cache line,
which can cause retries when there are concurrent
read and writes in one cache line.

Tested with IPoIB on PPC64, saw ~5% throughput
improvement.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/mlx4: Fix create CQ error flow

commit 593ff73bcfdc79f79a8a0df55504f75ad3e5d1a9 upstream.

Currently, if ib_copy_to_udata fails, the CQ
won't be deleted from the radix tree and the HW (HW2SW).

Fixes: 225c7b1feef1 ('IB/mlx4: Add a driver Mellanox ConnectX InfiniBand adapters')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/mlx4: Check gid_index return value

commit 37995116fecfce2b61ee3da6e73b3e394c6818f9 upstream.

Check the returned GID index value and return an error if it is invalid.

Fixes: 5070cd2239bd ('IB/mlx4: Replace mechanism for RoCE GID management')
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

PM / sleep: don't suspend parent when async child suspend_{noirq, late} fails

commit 6f75c3fd56daf547d684127a7f83c283c3c160d1 upstream.

Consider two devices, A and B, where B is a child of A, and B utilizes
asynchronous suspend (it does not matter whether A is sync or async). If
B fails to suspend_noirq() or suspend_late(), or is interrupted by a
wakeup (pm_wakeup_pending()), then it aborts and sets the async_error
variable. However, device A does not (immediately) check the async_error
variable; it may continue to run its own suspend_noirq()/suspend_late()
callback. This is bad.

We can resolve this problem by doing our error and wakeup checking
(particularly, for the async_error flag) after waiting for children to
suspend, instead of before. This also helps align the logic for the noirq and
late suspend cases with the logic in __device_suspend().

It's easy to observe this erroneous behavior by, for example, forcing a
device to sleep a bit in its suspend_noirq() (to ensure the parent is
waiting for the child to complete), then return an error, and watch the
parent suspend_noirq() still get called. (Or similarly, fake a wakeup
event at the right (or is it wrong?) time.)

Fixes: de377b397272 (PM / sleep: Asynchronous threads for suspend_late)
Fixes: 28b6fd6e3779 (PM / sleep: Asynchronous threads for suspend_noirq)
Reported-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

PM / sleep: fix device reference leak in test_suspend

commit ceb75787bc75d0a7b88519ab8a68067ac690f55a upstream.

Make sure to drop the reference taken by class_find_device() after
opening the RTC device.

Fixes: 77437fd4e61f (pm: boot time suspend selftest)
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

uwb: fix device reference leaks

commit d6124b409ca33c100170ffde51cd8dff761454a1 upstream.

This subsystem consistently fails to drop the device reference taken by
class_find_device().

Note that some of these lookup functions already take a reference to the
returned data, while others claim no reference is needed (or does not
seem need one).

Fixes: 183b9b592a62 ("uwb: add the UWB stack (core files)")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mfd: core: Fix device reference leak in mfd_clone_cell

commit 722f191080de641f023feaa7d5648caf377844f5 upstream.

Make sure to drop the reference taken by bus_find_device_by_name()
before returning from mfd_clone_cell().

Fixes: a9bbba996302 ("mfd: add platform_device sharing support for mfd")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iwlwifi: pcie: fix SPLC structure parsing

commit e0d9727c111a5917a1184c71c1a8e6f78c7fc41d upstream.

The SPLC data parsing is too restrictive and was not trying find the
correct element for WiFi.  This causes problems with some BIOSes where
the SPLC method exists, but doesn't have a WiFi entry on the first
element of the list.  The domain type values are also incorrect
according to the specification.

Fix this by complying with the actual specification.

Additionally, replace all occurrences of SPLX to SPLC, since SPLX is
only a structure internal to the ACPI tables, and may not even exist.

Fixes: bcb079a14d75 ("iwlwifi: pcie: retrieve and parse ACPI power limitations")
Reported-by: Chris Rorvick <chris@rorvick.com>
Tested-by: Paul Bolle <pebolle@tiscali.nl>
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

rtc: omap: Fix selecting external osc

commit 3984903a2e3906d3def220e688040ce93368200a upstream.

RTC can be clocked from an external 32KHz oscillator, or from the
Peripheral PLL. The RTC has an internal oscillator buffer to support
direct operation with a crystal.

            ----------------------------------------
            |       Device          ---------       |
            |                       |       |       |
            |                       | RTCSS |       |
            |       ---------       |       |       |
    OSC     |<------| RTC   |       |       |       |
            |------>| OSC   |---    |       |       |
            |       --------   |    |       |       |
            |                   ----|clk    |       |
            |       --------   |    |       |       |
            |       | PRCM  |---    |       |       |
            |       --------        --------        |
            ----------------------------------------

The RTC functional clock is sourced by default from the clock derived
from the Peripheral PLL. In order to select source as external osc clk
the following changes needs to be done:
- Enable the RTC OSC (RTC_OSC_REG[4]OSC32K_GZ = 0)
- Enable the clock mux(RTC_OSC_REG[6]K32CLK_EN = 1)
- Select the external clock source (RTC_OSC_REG[3]32KCLK_SEL = 1)

Fixes: 399cf0f63f6f2 ("rtc: omap: Add external clock enabling support")
Signed-off-by: Keerthy <j-keerthy@ti.com>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: mmp: mmp2: fix return value check in mmp2_clk_init()

commit a29e52a6e66f4c0c895e7083e4bad2e7957f1fb5 upstream.

Fix the retrn value check which testing the wrong variable
in mmp2_clk_init().

Fixes: 1ec770d92a62 ("clk: mmp: add mmp2 DT support for clock driver")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: mmp: pxa168: fix return value check in pxa168_clk_init()

commit deab07261d54b4db7b627d38e0efac97f176c6d6 upstream.

Fix the retrn value check which testing the wrong variable
in pxa168_clk_init().

Fixes: ab08aefcd12d ("clk: mmp: add pxa168 DT support for clock driver")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: mmp: pxa910: fix return value check in pxa910_clk_init()

commit 10f2bfb092e3b49000526c02cfe8b2abbbdbb752 upstream.

Fix the retrn value check which testing the wrong variable
in pxa910_clk_init().

Fixes: 2bc61da9f7ff ("clk: mmp: add pxa910 DT support for clock driver")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu: Attach exclusive fence to prime exported bo's. (v5)

commit 8e94a46c1770884166b31adc99eba7da65a446a7 upstream.

External clients which import our bo's wait only
for exclusive dmabuf-fences, not on shared ones,
ditto for bo's which we import from external
providers and write to.

Therefore attach exclusive fences on prime shared buffers
if our exported buffer gets imported by an external
client, or if we import a buffer from an external
exporter.

See discussion in thread:
https://lists.freedesktop.org/archives/dri-devel/2016-October/122370.html

Prime export tested on Intel iGPU + AMD Tonga dGPU as
DRI3/Present Prime render offload, and with the Tonga
standalone as primary gpu.

v2: Add a wait for all shared fences before prime export,
    as suggested by Christian Koenig.

v3: - Mark buffer prime_exported in amdgpu_gem_prime_pin,
    so we only use the exclusive fence when exporting a
    bo to external clients like a separate iGPU, but not
    when exporting/importing from/to ourselves as part of
    regular DRI3 fd passing.

    - Propagate failure of reservation_object_wait_rcu back
    to caller.

v4: - Switch to a prime_shared_count counter instead of a
      flag, which gets in/decremented on prime_pin/unpin, so
      we can switch back to shared fences if all clients
      detach from our exported bo.

    - Also switch to exclusive fence for prime imported bo's.

v5: - Drop lret, instead use int ret -> long ret, as proposed
      by Christian.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95472
Tested-by: Mike Lothian <mike@fireburn.co.uk> (v1)
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>.
Cc: Christian König <christian.koenig@amd.com>
Cc: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

crypto: caam - do not register AES-XTS mode on LP units

commit 83d2c9a9c17b1e9f23a3a0c24c03cd18e4b02520 upstream.

When using AES-XTS on a Wandboard, we receive a Mode error:
caam_jr 2102000.jr1: 20001311: CCB: desc idx 19: AES: Mode error.

According to the Security Reference Manual, the Low Power AES units
of the i.MX6 do not support the XTS mode. Therefore we must not
register XTS implementations in the Crypto API.

Signed-off-by: Sven Ebenfeld <sven.ebenfeld@gmail.com>
Reviewed-by: Horia Geantă <horia.geanta@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixes: c6415a6016bf "crypto: caam - add support for acipher xts(aes)"
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

ext4: sanity check the block and cluster size at mount time

commit 8cdf3372fe8368f56315e66bea9f35053c418093 upstream.

If the block size or cluster size is insane, reject the mount.  This
is important for security reasons (although we shouldn't be just
depending on this check).

Ref: http://www.securityfocus.com/archive/1/539661
Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1332506
Reported-by: Borislav Petkov <bp@alien8.de>
Reported-by: Nikolay Borisov <kernel@kyup.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

kbuild: Steal gcc's pie from the very beginning

commit c6a385539175ebc603da53aafb7753d39089f32e upstream.

So Sebastian turned off the PIE for kernel builds but that was too late
- Kbuild.include already uses KBUILD_CFLAGS and trying to disable gcc
options with, say cc-disable-warning, fails:

  gcc -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs
  ...
  -Wno-sign-compare -fno-asynchronous-unwind-tables -Wframe-address -c -x c /dev/null -o .31392.tmp
  /dev/null:1:0: error: code model kernel does not support PIC mode

because that returns an error and we can't disable the warning. For
example in this case:

KBUILD_CFLAGS   += $(call cc-disable-warning,frame-address,)

which leads to gcc issuing all those warnings again.

So let's turn off PIE/PIC at the earliest possible moment, when we
declare KBUILD_CFLAGS so that cc-disable-warning picks it up too.

Also, we need the $(call cc-option ...) because -fno-PIE is supported
since gcc v3.4 and our lowest supported gcc version is 3.2 right now.

Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/kexec: add -fno-PIE

commit 90944e40ba1838de4b2a9290cf273f9d76bd3bdd upstream.

If the gcc is configured to do -fPIE by default then the build aborts
later with:
| Unsupported relocation type: unknown type rel type name (29)

Tagging it stable so it is possible to compile recent stable kernels as
well.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scripts/has-stack-protector: add -fno-PIE

commit 82031ea29e454b574bc6f49a33683a693ca5d907 upstream.

Adding -no-PIE to the fstack protector check. -no-PIE was introduced
before -fstack-protector so there is no need for a runtime check.

Without it the build stops:
|Cannot use CONFIG_CC_STACKPROTECTOR_STRONG: -fstack-protector-strong available but compiler is broken

due to -mcmodel=kernel + -fPIE if -fPIE is enabled by default.

Tagging it stable so it is possible to compile recent stable kernels as
well.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

kbuild: add -fno-PIE

commit 8ae94224c9d72fc4d9aaac93b2d7833cf46d7141 upstream.

Debian started to build the gcc with -fPIE by default so the kernel
build ends before it starts properly with:
|kernel/bounds.c:1:0: error: code model kernel does not support PIC mode

Also add to KBUILD_AFLAGS due to:

|gcc -Wp,-MD,arch/x86/entry/vdso/vdso32/.note.o.d … -mfentry -DCC_USING_FENTRY … vdso/vdso32/note.S
|arch/x86/entry/vdso/vdso32/note.S:1:0: sorry, unimplemented: -mfentry isn’t supported for 32-bit in combination with -fpic

Tagging it stable so it is possible to compile recent stable kernels as
well.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

i2c: mux: fix up dependencies

commit 93d710a65ef02fb7fd48ae207e78f460bd7a6089 upstream.

We get the following build error from UM Linux after adding
an entry to drivers/iio/gyro/Kconfig that issues "select I2C_MUX":

ERROR: "devm_ioremap_resource"
   [drivers/i2c/muxes/i2c-mux-reg.ko] undefined!
ERROR: "of_address_to_resource"
   [drivers/i2c/muxes/i2c-mux-reg.ko] undefined!

It appears that the I2C mux core code depends on HAS_IOMEM
for historical reasons, while CONFIG_I2C_MUX_REG does *not*
have a direct dependency on HAS_IOMEM.

This creates a situation where a allyesconfig or allmodconfig
for UM Linux will select I2C_MUX, and will implicitly enable
I2C_MUX_REG as well, and the compilation will fail for the
register driver.

Fix this up by making I2C_MUX_REG depend on HAS_IOMEM and
removing the dependency from I2C_MUX.

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Reported-by: Jonathan Cameron <jic23@jic23.retrosnub.co.uk>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jonathan Cameron <jic23@kernel.org>
Acked-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

can: bcm: fix warning in bcm_connect/proc_register

commit deb507f91f1adbf64317ad24ac46c56eeccfb754 upstream.

Andrey Konovalov reported an issue with proc_register in bcm.c.
As suggested by Cong Wang this patch adds a lock_sock() protection and
a check for unsuccessful proc_create_data() in bcm_connect().

Reference: http://marc.info/?l=linux-netdev&m=147732648731237

Reported-by: Andrey Konovalov <andreyknvl@google.com>
Suggested-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Tested-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>