arm64: rockchip: rk3368 support psci-0.2

Signed-off-by: Huang, Tao <huangtao@rock-chips.com>

arm64: rockchip: rk3368 fix qos reg define and increase peri priority

Signed-off-by: Huang, Tao <huangtao@rock-chips.com>

rk3368: support cpu suspend to wfi

Signed-off-by: chenjh <chenjh@rock-chips.com>

IEP: revise the dtsi clock gate number.

Incorrect clock gate provided in previous commit.

Signed-off-by: Alpha Lin <alpha.lin@rock-chips.com>

rk3x:pmic:rk818:not init rk818_ldo3 in uboot

rk fb: update pixclock init value and scale mode only support ONE_DUAL mode

Signed-off-by: hjc <hjc@rock-chips.com>

rk3368-p9: dts: change rk818 battery configs.

regmap: Make regmap-mmio usable from atomic contexts

regmap-mmio uses a spinlock with spin_lock() and spin_unlock() for locking.
To be able to use the regmap API from different contexts (atomic vs non-atomic),
without the risk of race conditions, we need to use spin_lock_irqsave() and
spin_lock_irqrestore() instead. A new field, the spinlock_flags field, is added
to regmap struct to store the flags between regmap_{,un}lock_spinlock(). The
spinlock_flags field itself is also protected by the spinlock.

Thanks to Stephen Warren for the suggestion of this particular solution.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
(cherry picked from commit 92ab1aab59c61b3e05200b9aa0e05ab770059142)

arm64: rockchip: rk3368 support boot from little core

Signed-off-by: Huang, Tao <huangtao@rock-chips.com>

rk3368 hdmi: support pd_hdmi.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

Merge branch develop-3.10

rk fb: compatible with 32 bit system

Signed-off-by: hjc <hjc@rock-chips.com>

rk3368 lcdc: add yuv420sp nv21 format at lcdc scale fac

Signed-off-by: hjc <hjc@rock-chips.com>

rga2 support 32bit compile at 64bit platform

fix conpile error

wifi: use buildin instand of ko for compatibility

mmc: rk_sdmmc: add uart & jtag soft-switch by routine card present signal

Reviewed-and-tested-by: <xiaoyao@rock-chips.com>
Signed-off-by: lintao <lintao@rock-chips.com>

dtsi: rk3368: add uart & jtag pcl func for sdmmc mux

Signed-off-by: lintao <lintao@rock-chips.com>

VPU: revise previous commit.

can not be built in previous commit.

Signed-off-by: Alpha Lin <alpha.lin@rock-chips.com>

dts: add work and power led in rk3128-box.dts

Merge branch 'develop-3.10' of ssh://10.10.10.29/rk/kernel into develop-3.10

VPU, enable frequency raise for 4k h.264 decoding.

Signed-off-by: Alpha Lin <alpha.lin@rock-chips.com>

arm64: add PSCI CPU_SUSPEND based cpu_suspend support

This patch implements the cpu_suspend cpu operations method through
the PSCI CPU SUSPEND API. The PSCI implementation translates the idle state
index passed by the cpu_suspend core call into a valid PSCI state according to
the PSCI states initialized at boot through the cpu_init_idle() CPU
operations hook.

The PSCI CPU suspend operation hook checks if the PSCI state is a
standby state. If it is, it calls the PSCI suspend implementation
straight away, without saving any context. If the state is a power
down state the kernel calls the __cpu_suspend API (that saves the CPU
context) and passed the PSCI suspend finisher as a parameter so that PSCI
can be called by the __cpu_suspend implementation after saving and flushing
the context as last function before power down.

For power down states, entry point is set to cpu_resume physical address,
that represents the default kernel execution address following a CPU reset.

Reviewed-by: Ashwin Chaugule <ashwin.chaugule@linaro.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
(cherry picked from commit 18910ab0d916b1a87016d69efd027714a80521dd)
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
Conflicts:
arch/arm64/kernel/psci.c

arm64: kernel: fix __cpu_suspend mm switch on warm-boot

On arm64 the TTBR0_EL1 register is set to either the reserved TTBR0
page tables on boot or to the active_mm mappings belonging to user space
processes, it must never be set to swapper_pg_dir page tables mappings.

When a CPU is booted its active_mm is set to init_mm even though its
TTBR0_EL1 points at the reserved TTBR0 page mappings. This implies
that when __cpu_suspend is triggered the active_mm can point at
init_mm even if the current TTBR0_EL1 register contains the reserved
TTBR0_EL1 mappings.

Therefore, the mm save and restore executed in __cpu_suspend might
turn out to be erroneous in that, if the current->active_mm corresponds
to init_mm, on resume from low power it ends up restoring in the
TTBR0_EL1 the init_mm mappings that are global and can cause speculation
of TLB entries which end up being propagated to user space.

This patch fixes the issue by checking the active_mm pointer before
restoring the TTBR0 mappings. If the current active_mm == &init_mm,
the code sets the TTBR0_EL1 to the reserved TTBR0 mapping instead of
switching back to the active_mm, which is the expected behaviour
corresponding to the TTBR0_EL1 settings when __cpu_suspend was entered.

Fixes: 95322526ef62 ("arm64: kernel: cpu_{suspend/resume} implementation")
Cc: <stable@vger.kernel.org> # 3.14+: 18ab7db
Cc: <stable@vger.kernel.org> # 3.14+: 714f599
Cc: <stable@vger.kernel.org> # 3.14+: c3684fb
Cc: <stable@vger.kernel.org> # 3.14+
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
(cherry picked from commit f43c27188a49111b58e9611afa2f0365b0b55625)

arm64: Move cpu_resume into the text section

The function cpu_resume currently lives in the .data section.
There's no reason for it to be there since we can use relative
instructions without a problem. Move a few cpu_resume data
structures out of the assembly file so the .data annotation
can be dropped completely and cpu_resume ends up in the read
only text section.

Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Kees Cook <keescook@chromium.org>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
(cherry picked from commit c3684fbb446501b48dec6677a6a9f61c215053de)

arm64: kernel: refactor the CPU suspend API for retention states

CPU suspend is the standard kernel interface to be used to enter
low-power states on ARM64 systems. Current cpu_suspend implementation
by default assumes that all low power states are losing the CPU context,
so the CPU registers must be saved and cleaned to DRAM upon state
entry. Furthermore, the current cpu_suspend() implementation assumes
that if the CPU suspend back-end method returns when called, this has
to be considered an error regardless of the return code (which can be
successful) since the CPU was not expected to return from a code path that
is different from cpu_resume code path - eg returning from the reset vector.

All in all this means that the current API does not cope well with low-power
states that preserve the CPU context when entered (ie retention states),
since first of all the context is saved for nothing on state entry for
those states and a successful state entry can return as a normal function
return, which is considered an error by the current CPU suspend
implementation.

This patch refactors the cpu_suspend() API so that it can be split in
two separate functionalities. The arm64 cpu_suspend API just provides
a wrapper around CPU suspend operation hook. A new function is
introduced (for architecture code use only) for states that require
context saving upon entry:

__cpu_suspend(unsigned long arg, int (*fn)(unsigned long))

__cpu_suspend() saves the context on function entry and calls the
so called suspend finisher (ie fn) to complete the suspend operation.
The finisher is not expected to return, unless it fails in which case
the error is propagated back to the __cpu_suspend caller.

The API refactoring results in the following pseudo code call sequence for a
suspending CPU, when triggered from a kernel subsystem:

/*
 * int cpu_suspend(unsigned long idx)
 * @idx: idle state index
 */
{
-> cpu_suspend(idx)
|---> CPU operations suspend hook called, if present
|--> if (retention_state)
|--> direct suspend back-end call (eg PSCI suspend)
     else
|--> __cpu_suspend(idx, &back_end_finisher);
}

By refactoring the cpu_suspend API this way, the CPU operations back-end
has a chance to detect whether idle states require state saving or not
and can call the required suspend operations accordingly either through
simple function call or indirectly through __cpu_suspend() which carries out
state saving and suspend finisher dispatching to complete idle state entry.

Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Hanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
(cherry picked from commit 714f59925595b9c2ea9c22b107b340d38e3b3bc9)

arm64: kernel: add missing __init section marker to cpu_suspend_init

Suspend init function must be marked as __init, since it is not needed
after the kernel has booted. This patch moves the cpu_suspend_init()
function to the __init section.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
(cherry picked from commit 18ab7db6b749ac27aac08d572afbbd2f4d937934)

driver/core: cpu: initialize of_node in cpu's device struture

CPUs are also registered as devices but the of_node in these cpu
devices are not initialized. Currently different drivers requiring
to access cpu device node are parsing the nodes themselves and
initialising the of_node in cpu device.

The of_node in all the cpu devices needs to be initialized properly
and at one place. The best place to update this is CPU subsystem
driver when registering the cpu devices.

The OF/DT core library now provides of_get_cpu_node to retrieve a cpu
device node for a given logical index by abstracting the architecture
specific details.

This patch uses of_get_cpu_node to assign of_node when registering the
cpu devices.

Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Sudeep KarkadaNagesha <sudeep.karkadanagesha@arm.com>
(cherry picked from commit f86e4718f24b83be0c42894d2b97accc993d65d3)
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
Conflicts:
drivers/base/cpu.c

Merge branch 'develop-3.10-next' of ssh://10.10.10.29/rk/kernel into develop-3.10-next

Merge branch linux-linaro-lsk-v3.10-android

dts: update rk3368-box dts:1, wifi_bt,2,enabled pwm1;3.emmc clk

pd: rk3368: add rk3368 power domain support (as pd clk)

Signed-off-by: dkl <dkl@rock-chips.com>

Merge branch 'develop-3.10' of ssh://10.10.10.29/rk/kernel into develop-3.10

usb: dwc_otg: fix bug in device_init routine

1.wait epenable bit timeout decrease from 10s to 1s
2.fix bug, without 'j++' this loop never break if there is some
problem with ep state

dts: rk3368: remove "clk_ignore_unused" from bootargs

Signed-off-by: dkl <dkl@rock-chips.com>

usb: ehci: fix suspend bug

usb: dwc_otg: change method of checking urb dequeue status

According to commit e9df41c5c, change dwc_otg driver to call
usb_hcd_link_urb_to_ep() and usb_hcd_unlink_urb_from_ep() routine
for checking urb dequeue status instead check for null pointer
by hcd driver to test address.

TEST:
1.rk3126 run host/device change script for 10000 times

rk818-battery: add calibration of charge and discharge

Signed-off-by: 许盛飞 <xsf@rock-chips.com>

rk818-battery: enable parse battery params

Signed-off-by: 许盛飞 <xsf@rock-chips.com>

rk818-battery: reconstruct the voltage discharge algorithm

Signed-off-by: 许盛飞 <xsf@rock-chips.com>

rk3368-box.dts: set initial display timing to 1080P.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

rk32xx lvds: add support lvds 10bit mode

Signed-off-by: hjc <hjc@rock-chips.com>

3368_P9: dts: renamed 818 DCDC2 regulator to vdd_logic

Merge branch 'develop-3.10' of ssh://10.10.10.29/rk/kernel into develop-3.10

after support different wifi bt chip auto compatible commit for android 5.0, compatible with android 4.4

VPU: remove dev_set_name.

dev_set_name is prohibit in linux, and may casue system
warning dump.

Signed-off-by: Alpha Lin <alpha.lin@rock-chips.com>

usb: fix compliation warning without CONFIG_RK_DEBUG_UART

rk3288:tsadc:enable tsadc chn3

rk312x lcdc: add to get hwc layer status

Signed-off-by: hjc <hjc@rock-chips.com>

bmp_helper: fix bmp rle8 display abnormal

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

rk312x lcdc: add to get hwc layer status

Signed-off-by: hjc <hjc@rock-chips.com>

rtc: HYM8563: don't call rtc_year_days

If i2c error, the driver may call rtc_year_days with month -1, which will
cause crash. Fix this by never call rtc_year_days at all.

Signed-off-by: Huang, Tao <huangtao@rock-chips.com>

battery: enable dc charge

Signed-off-by: 许盛飞 <xsf@rock-chips.com>

update rtl8188eu wifi driver to v4.3.0.8_13522.20150213_beta
 solve android 5.0 ctsverify wifi direct test fail problem

bmp_helper: fix bmp rle8 display abnormal

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

ESP8089 sdio premalloc default config

wifi: support ap6212 ap6354

arm64: ion: fix scatterlist alloc fault on ion_system heap

Fix this bug:
WARNING: at lib/scatterlist.c:251 __sg_alloc_table+0x5c/0x15c()
Modules linked in: pvrsrvkm(O)
CPU: 2 PID: 582 Comm: Binder_5 Tainted: G           O 3.10.0 #10
Call trace:
[<ffffffc000088164>] dump_backtrace+0x0/0x140
[<ffffffc0000882b4>] show_stack+0x10/0x1c
[<ffffffc0008ccaf0>] dump_stack+0x1c/0x28
[<ffffffc0000a13c0>] warn_slowpath_common+0x74/0x9c
[<ffffffc0000a14ac>] warn_slowpath_null+0x14/0x20
[<ffffffc0002fadc4>] __sg_alloc_table+0x58/0x15c
[<ffffffc0002faee8>] sg_alloc_table+0x20/0x50
[<ffffffc00067ab7c>] ion_system_heap_allocate+0x188/0x318
[<ffffffc00067823c>] ion_alloc+0x134/0x788
[<ffffffc000678d68>] ion_ioctl+0x15c/0x344
[<ffffffc0001840ac>] vfs_ioctl+0x20/0x3c
[<ffffffc0001849b4>] do_vfs_ioctl+0x478/0x550
[<ffffffc000184ae8>] SyS_ioctl+0x5c/0x88

Signed-off-by: CMY <cmy@rock-chips.com>

rk3368 dtsi: update vop iommu compatible name

Signed-off-by: hjc <hjc@rock-chips.com>

Update RTL8723BS wifi driver
1.Refine IQK handle
2.Fix RX error handle to avoid system hang problem
3.Fix IQK restore fail issue
4.Fix running out of RX resource in some case
5.Fix HiddenAP still in driver scan queue after disconnect

Espressi esp8089 WIFI
1.modify copyright info for ESP8089 WIFI driver
2.update esp_premalloc driver to V2.5

Merge branch 'linux-linaro-lsk' into linux-linaro-lsk-android

Merge remote-tracking branch 'lsk/v3.10/topic/cpufreq' into linux-linaro-lsk

cpufreq: Revert "cpufreq: suspend governors on system suspend/hibernate"

This reverts commit adb665056afe23db0687c1d1a208806c108add08.

The commit has too many hidden dependencies to be sensibly backported.

Signed-off-by: Mark Brown <broonie@linaro.org>

cpufreq: Revert "cpufreq: Don't iterate over all CPUs when suspending and resuming governors"

This reverts commit 212df5dd3212092602119d03a2f5d3a27be77ce4.

The commit it's fixing has too many other dependencies.

Signed-off-by: Mark Brown <broonie@linaro.org>

rk3368 lcdc: FBDC_ABGR_888 data format support alpha

Signed-off-by: hjc <hjc@rock-chips.com>

rk fb: if not want the config effect,set reserved[3] bit[0] 1

Signed-off-by: hjc <hjc@rock-chips.com>

update wifi bcmdhd driver:
 solve kernel panic when use gmac and hotspots at same time

update wifi bcmdhd driver:
 solve kernel panic when use gmac and hotspots at same time

rk fb: add fbdc data format: HAL_PIXEL_FORMAT_FBDC_RGBA888

dts: rk3036/rk312x/rk3288: add dma clk to fit
commit 75ce2b93307105d10a21e995ab27ec30cc50aaf3

Signed-off-by: dkl <dkl@rock-chips.com>

rk3368 lcdc: add fbdc data format: HAL_PIXEL_FORMAT_FBDC_RGBA888

rk3368 lcdc:
    1. check act and size in win2 win3 layer;
    2. add support yuv420sp nv21 data format;
    3. in read bit func,not need to backup reg;
    4. complete get win state func;

Signed-off-by: hjc <hjc@rock-chips.com>

rk fb: add memsize check

Signed-off-by: hjc <hjc@rock-chips.com>

rk lcdc: if win not support scale, make sure dsp size equal to act size

Signed-off-by: hjc <hjc@rock-chips.com>

rk fb: add support yuv420sp nv21 data format

Signed-off-by: hjc <hjc@rock-chips.com>
Conflicts:
drivers/video/rockchip/rkfb_sysfs.c
include/linux/rk_fb.h

rk fb: make sure win close state take effect when update regs

Signed-off-by: hjc <hjc@rock-chips.com>

rk3368: clk: modify relation in software of clocks under PD_VIO

It is necessary to enable hclk_vio_h2p\pclk_vio_h2p\hclk_vio_ahb_arbi
for devices under power doamin PD_VIO wen they work. Thus these clks
are modified to be parent of other clks from hclk_vio.

Signed-off-by: dkl <dkl@rock-chips.com>

DMA: rk3368: add dma clk enable/disable

Signed-off-by: dkl <dkl@rock-chips.com>

rk fb: define data format for fbdc

Signed-off-by: hjc <hjc@rock-chips.com>
Conflicts:
drivers/video/rockchip/lcdc/rk3368_lcdc.c
drivers/video/rockchip/rkfb_sysfs.c
include/linux/rk_fb.h

rk fb: add memsize check

Signed-off-by: hjc <hjc@rock-chips.com>

rk lcdc: if win not support scale, make sure dsp size equal to act size

Signed-off-by: hjc <hjc@rock-chips.com>

rk fb: add support yuv420sp nv21 data format

Signed-off-by: hjc <hjc@rock-chips.com>

rk fb: make sure win close state take effect when update regs

Signed-off-by: hjc <hjc@rock-chips.com>

HDMI: Modify color output policy.
        Introduce feature SUPPORT_YCBCR_INPUT. If VOP can
        output YCbCr, VOP output color mode is equal to HDMI
        output color mode. Otherwise, VOP just output RGB,
        HDMI will conver color space.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

HDMI: add mutex lock when hdmi is set to auto mode, to avoid
      system crash by NULL pointer.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

HDMI: Modify color output policy.
        Introduce feature SUPPORT_YCBCR_INPUT. If VOP can
        output YCbCr, VOP output color mode is equal to HDMI
        output color mode. Otherwise, VOP just output RGB,
        HDMI will conver color space.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

HDMI: add mutex lock when hdmi is set to auto mode, to avoid
      system crash by NULL pointer.

Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>

solve bluetooth hid problem:
 when bluetooth hid device connected, the key is delayed to be reponse about 5 second

rk3368: thermal: add log trigger for tsadc
enable log:
echo t > sys/rockchip_thermal/temp
disable log:
echo f > sys/rockchip_thermal/temp

Signed-off-by: David Wu <wdc@rock-chips.com>

rk3368: tb_8846: change 8846 mode for tsadc

Signed-off-by: David Wu <wdc@rock-chips.com>

rk3368: add tsadc for p9

Signed-off-by: David Wu <wdc@rock-chips.com>

rk3368: tsadc: remove tsadc-int pin for p9

Signed-off-by: David Wu <wdc@rock-chips.com>

ethernet: fix dts for rk3368-box

Signed-off-by: chenzhi <roger.chen@rock-chips.com>

rk3368: thermal: remove dbg log

Signed-off-by: David Wu <wdc@rock-chips.com>

rk3368: use pvtm for 32k tep

Signed-off-by: David Wu <wdc@rock-chips.com>

rk3368: tsadc: remove tsadc_tockchip defconfig

Signed-off-by: David Wu <wdc@rock-chips.com>

usb: can be wakeup by linestate change interrupt from GRF

rk3368: thermal: tsadc use usermode for rk3368

Signed-off-by: David Wu <wdc@rock-chips.com>

pvtm: rk3368: add pvtm_pmu clk support (output 32k)

Support pvtm_pmu output 32k clk. By default, 32k use xin32k,
and pvtm_pmu 32k clk is disabled.

Signed-off-by: dkl <dkl@rock-chips.com>