Merge tag 'samsung-drivers-2' of git://git./linux/kernel/git/kgene/linux-samsung into next/drivers

Merge "Samsung 2nd drivers for 3.16" from Kukjin Kim:

This is including fix exynos cpufreq driver compilation with
ARCH_MULTIPLATFORM. Even though this is a work around, this
is required for support exynos multiplatform for a while and
will be updated in near future.

This is based on tags/samsung-exynos.

* tag 'samsung-drivers-2' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung: (24 commits)
  cpufreq: exynos: Fix driver compilation with ARCH_MULTIPLATFORM
  ARM: EXYNOS: Enable multi-platform build support
  ARM: EXYNOS: Consolidate Kconfig entries
  ARM: EXYNOS: Add support for EXYNOS5410 SoC
  ARM: EXYNOS: Support secondary CPU boot of Exynos3250
  ARM: EXYNOS: Add Exynos3250 SoC ID
  ARM: EXYNOS: Add 5800 SoC support
  ARM: EXYNOS: initial board support for exynos5260 SoC
  clk: exynos5250: Add missing sysmmu clocks for DISP and ISP blocks
  cpufreq: exynos: Fix the compile error
  ARM: S3C24XX: move debug-macro.S into the common space
  ARM: S3C24XX: use generic DEBUG_UART_PHY/_VIRT in debug macro
  ARM: S3C24XX: trim down debug uart handling
  ARM: compressed/head.S: remove s3c24xx special case
  ARM: EXYNOS: Remove unnecessary inclusion of cpu.h
  ARM: EXYNOS: Migrate Exynos specific macros from plat to mach
  ARM: EXYNOS: Remove exynos_subsys registration
  ARM: EXYNOS: Remove duplicate lines in Makefile
  ARM: EXYNOS: use v7_exit_coherency_flush macro for cache disabling
  ARM: dts: Remove g2d_pd node for exynos5420
  ...

Signed-off-by: Olof Johansson <[email protected]>

Merge tag 'qcom-drivers-for-3.16-2' of git://git./linux/kernel/git/galak/linux-qcom into next/drivers

Merge "Qualcomm ARM Based driver Updates for v3.16-2" from Kumar Gala:

* Updated MSM uart driver to not access GSBI control registers

* tag 'qcom-drivers-for-3.16-2' of git://git.kernel.org/pub/scm/linux/kernel/git/galak/linux-qcom:
  tty: serial: msm: Remove direct access to GSBI

Signed-off-by: Olof Johansson <[email protected]>

cpufreq: exynos: Fix driver compilation with ARCH_MULTIPLATFORM

Currently Exynos cpufreq drivers rely on globally mapped
clock controller registers to configure frequency of CPU
cores. This is obviously wrong and will be removed in near
future, but to enable support for multi-platform builds
without introducing a regression it needs to be worked
around.

This patch hacks the code to look for clock controller node
in device tree and map its registers using of_iomap(),
instead of relying on global mapping, so dependencies on
platform headers are removed and the driver can compile
again with multiplatform support.

Signed-off-by: Tomasz Figa <[email protected]>
Acked-by: Viresh Kumar <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Enable multi-platform build support

This makes it possible to enable the Exynos platform as part of a
multiplatform kernel.

Signed-off-by: Arnd Bergmann <[email protected]>
Signed-off-by: Sachin Kamat <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Consolidate Kconfig entries

Instead of repeating the Kconfig entries for every SoC,
move them under ARCH_EXYNOS3, 4 and 5 and move the entries
common to 3, 4 and 5 under ARCH_EXYNOS.

Signed-off-by: Sachin Kamat <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add support for EXYNOS5410 SoC

EXYNOS5410 is SoC in Samsung's Exynos5 SoC series.
Add initial support for this SoC.

Signed-off-by: Tarek Dakhran <[email protected]>
Signed-off-by: Vyacheslav Tyrtov <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Support secondary CPU boot of Exynos3250

This patch fix the offset of CPU boot address and don't
need to send smc call of SMC_CMD_CPU1BOOT command for
secondary CPU boot because Exynos3250 removes WFE in
secure mode.

Signed-off-by: Chanwoo Choi <[email protected]>
Acked-by: Kyungmin Park <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add Exynos3250 SoC ID

This patch add Exynos3250's SoC ID. Exynos 3250 is SoC that
is based on the 32-bit RISC processor for Smartphone.
Exynos3250 uses Cortex-A7 dual cores and has a target speed
of 1.0GHz.

Signed-off-by: Chanwoo Choi <[email protected]>
Acked-by: Kyungmin Park <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add 5800 SoC support

Exynos5800 is an octa core SoC which is based on the 5420
platform. This patch adds the basic support for it in the
mach-exynos.

Signed-off-by: Arun Kumar K <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: initial board support for exynos5260 SoC

This patch add basic arch side support for exynos5260 SoC.
Note that this is required to enable build for clock driver.

Signed-off-by: Pankaj Dubey <[email protected]>
Signed-off-by: Rahul Sharma <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'v3.16-next/cleanup-samsung' into v3.16-next/platform-exynos

Merge tag 'exynos-cpuidle' of git./linux/kernel/git/kgene/linux-samsung into next/drivers

Merge "Samsung exynos-cpuidle updates for v3.16" from Kukjin Kim:

- From Daniel Lezcano:
 This patchset relies on the cpm_pm notifier to initiate the
 powerdown sequence operations from pm.c instead cpuidle.c.
 Thus the cpuidle driver is no longer dependent from arch
 specific code as everything is called from the pm.c file.

* tag 'exynos-cpuidle' of http://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung: (94 commits)
  ARM: EXYNOS: Fix kernel panic when unplugging CPU1 on exynos
  ARM: EXYNOS: Move the driver to drivers/cpuidle directory
  ARM: EXYNOS: Cleanup all unneeded headers from cpuidle.c
  ARM: EXYNOS: Pass the AFTR callback to the platform_data
  ARM: EXYNOS: Move S5P_CHECK_SLEEP into pm.c
  ARM: EXYNOS: Move the power sequence call in the cpu_pm notifier
  ARM: EXYNOS: Move the AFTR state function into pm.c
  ARM: EXYNOS: Encapsulate the AFTR code into a function
  ARM: EXYNOS: Disable cpuidle for exynos5440
  ARM: EXYNOS: Encapsulate boot vector code into a function for cpuidle
  ARM: EXYNOS: Pass wakeup mask parameter to function for cpuidle
  ARM: EXYNOS: Remove ifdef for scu_enable in pm
  ARM: EXYNOS: Move scu_enable in the cpu_pm notifier
  ARM: EXYNOS: Use the cpu_pm notifier for pm
  ARM: EXYNOS: Fix S5P_WAKEUP_STAT call for cpuidle
  ARM: EXYNOS: Move some code inside the idle_finisher for cpuidle
  ARM: EXYNOS: Encapsulate register access inside a function for pm
  ARM: EXYNOS: Change function name prefix for cpuidle
  ARM: EXYNOS: Use cpuidle_register
  ARM: EXYNOS: Prevent forward declaration for cpuidle
  ...

Signed-off-by: Olof Johansson <[email protected]>

tty: serial: msm: Remove direct access to GSBI

This patch removes direct access of the GSBI registers.  GSBI configuration
should be done through the GSBI driver directly.

Signed-off-by: Andy Gross <[email protected]>
Signed-off-by: Kumar Gala <[email protected]>

Merge tag 'keystone-reset-driver' of git://git./linux/kernel/git/ssantosh/linux-keystone into next/drivers

Merge "Keystone Reset driver for 3.16" from Santosh Shilimkar:

* tag 'keystone-reset-driver' of git://git.kernel.org/pub/scm/linux/kernel/git/ssantosh/linux-keystone:
  power: reset: keystone-reset: introduce keystone reset driver
  Documentation: dt: add bindings for keystone pll control controller
  Documentation: dt: add bindings for keystone reset driver

Signed-off-by: Olof Johansson <[email protected]>

power: reset: keystone-reset: introduce keystone reset driver

The keystone SoC can be rebooted in several ways. By external reset
pin, by soft and by watchdogs. To allow keystone SoC reset if
watchdog is triggered we have to enable it in reset mux configuration
register regarding of watchdog configuration. Also we need to set
soft/hard reset we are going to use.

So add keystone reset driver to handle all this stuff.

Reviewed-by: Arnd Bergmann <[email protected]>
Signed-off-by: Ivan Khoronzhuk <[email protected]>
Signed-off-by: Santosh Shilimkar <[email protected]>

Documentation: dt: add bindings for keystone pll control controller

The main pll controller used to drive theC66x CorePacs, the switch fabric,
and a majority of the peripheral clocks (all but the ARM CorePacs, DDR3 and
the NETCP modules) requires a PLL Controller to manage the various clock
divisions, gating, and synchronization.

Reviewed-by: Arnd Bergmann <[email protected]>
Acked-by: Mike Turquette <[email protected]>
Signed-off-by: Ivan Khoronzhuk <[email protected]>
[[email protected]: Fixed the subject line]
Signed-off-by: Santosh Shilimkar <[email protected]>

Documentation: dt: add bindings for keystone reset driver

This node is intended to allow SoC reset in case of software reset
or appropriate watchdogs.

The Keystone SoCs can contain up to 4 watchdog timers to reset
SoC. Each watchdog timer event input is connected to the Reset Mux
block. The Reset Mux block can be configured to cause reset or not.

Additionally soft or hard reset can be configured.

Reviewed-by: Arnd Bergmann <[email protected]>
Signed-off-by: Ivan Khoronzhuk <[email protected]>
[[email protected]: Fixed the subject line]
Signed-off-by: Santosh Shilimkar <[email protected]>

Merge tag 'davinci-for-v3.16/edma' of git://git./linux/kernel/git/nsekhar/linux-davinci into next/drivers

Merge "DaVinci EDMA clean-up for v3.16" from Sekhar Nori:

This series makes edma use configuration information available within
the IP instead of reading it from platform data or DT. Some other useful
clean-ups are included too.

* tag 'davinci-for-v3.16/edma' of git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci: (34 commits)
  ARM: edma: Remove redundant/unused parameters from edma_soc_info
  ARM: davinci: Remove redundant/unused parameters for edma
  ARM: dts: am4372: Remove obsolete properties from edma node
  ARM: dts: am33xx: Remove obsolete properties from edma node
  dt/bindings: ti,edma: Remove redundant properties from documentation
  ARM: edma: Get IP configuration from HW (number of channels, tc, etc)
  ARM: edma: Save number of regions from pdata to struct edma
  ARM: edma: Remove num_cc member from struct edma
  ARM: edma: Remove queue_tc_mapping data from edma_soc_info
  ARM: davinci: Remove eDMA3 queue_tc_mapping data from edma_soc_info
  ARM: edma: Do not change TC -> Queue mapping, leave it to default.
  ARM: edma: Take the number of tc from edma_soc_info (pdata)
  ARM: edma: No need to clean the pdata in edma_of_parse_dt()
  ARM: edma: Clean up and simplify the code around irq request
  dmaengine: edma: update DMA memcpy to use new param element
  dmaengine: edma: Document variables used for residue accounting
  dmaengine: edma: Provide granular accounting
  dmaengine: edma: Make reading the position of active channels work
  dmaengine: edma: Store transfer data in edma_desc and edma_pset
  dmaengine: edma: Create private pset struct
  ...

Signed-off-by: Olof Johansson <[email protected]>

Merge tag 'davinci-fixes-for-v3.15-rc4' of git://git./linux/kernel/git/nsekhar/linux-davinci into next/drivers

The patch fixes EDMA crossbar mapping to actually
make it work. The patch has been tagged for stable.

* tag 'davinci-fixes-for-v3.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci:
  ARM: common: edma: Fix xbar mapping

Merge tag 'omap-for-v3.16/l3-noc-signed' of git://git./linux/kernel/git/tmlind/linux-omap into next/drivers

Merge "ARM: omap l3-noc bus driver changes for v3.16 merge window, resend" from
Tony Lindgren:

Improvments to omap l3-noc bus driver for v3.16 merge window
to add support for am347x and dra7.

* tag 'omap-for-v3.16/l3-noc-signed' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap: (25 commits)
  bus: omap_l3_noc: Add AM4372 interconnect error data
  bus: omap_l3_noc: Add DRA7 interconnect error data
  bus: omap_l3_noc: introduce concept of submodule
  bus: omap_l3_noc: Add information about the context of operation
  bus: omap_l3_noc: add information about the type of operation
  bus: omap_l3_noc: ignore masked out unclearable targets
  bus: omap_l3_noc: improve readability by using helper for slave event parsing
  bus: omap_l3_noc: make error reporting and handling common
  bus: omap_l3_noc: fix masterid detection
  bus: omap_l3_noc: convert flagmux information into a structure
  bus: omap_l3_noc: use of_match_data to pick up SoC information
  bus: omap_l3_noc: Add support for discountinous flag mux input numbers
  bus: omap_l3_noc: convert target information into a structure
  bus: omap_l3_noc: move L3 master data structure out
  bus: omap_l3_noc: un-obfuscate l3_targ address computation
  bus: omap_l3_noc: switch over to relaxed variants of readl/writel
  bus: omap_l3_noc: populate l3->dev and use it
  bus: omap_l3_noc: remove iclk from omap_l3 struct
  bus: omap_l3_noc: rename functions and data to omap_l3
  bus: omap_l3_noc: Fix copyright information
  ...

Signed-off-by: Olof Johansson <[email protected]>

soc: qcom: fix of_device_id table

The match tables must be zero-terminated, and Kbuild now helpfully
fails to link the kernel if that isn't the case.

Signed-off-by: Arnd Bergmann <[email protected]>

clk: exynos5250: Add missing sysmmu clocks for DISP and ISP blocks

This patch adds the missing sysmmu clocks for Display and
ISP blocks.

Signed-off-by: Cho KyongHo <[email protected]>
Signed-off-by: Shaik Ameer Basha <[email protected]>
Acked-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Fix kernel panic when unplugging CPU1 on exynos

A look at the code reveals use of S5P_VA_SYSRAM macro, in case of certain SoC
revisions, which is not valid any longer, after SYSRAM started to be mapped
dynamically. The new dynamic mapping is stored in sysram_base_addr variable,
which is declared static in platsmp.c

This fix makes sysram_base_addr non-static, declared it in common.h and used
in pm.c instead of S5P_VA_SYSRAM.

Suggested-by: Tomasz Figa <[email protected]>
Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move the driver to drivers/cpuidle directory

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Cleanup all unneeded headers from cpuidle.c

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Pass the AFTR callback to the platform_data

No more dependency on the arch code. The platform_data field is used to set the
PM callback as the other cpuidle drivers.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move S5P_CHECK_SLEEP into pm.c

This macro is only used there.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move the power sequence call in the cpu_pm notifier

The code to initiate and exit the powerdown sequence is the same in
pm.c and cpuidle.c.

Let's split the common part in the pm.c and reuse it from the cpu_pm notifier.

That is one more step forward to make the cpuidle driver arch indenpendant.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move the AFTR state function into pm.c

In order to remove depedency on pm code, let's move the 'exynos_enter_aftr'
function into the pm.c file as well as the other helper functions.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Encapsulate the AFTR code into a function

Let's encapsulate the AFTR state specific call into a single function.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Disable cpuidle for exynos5440

There is no point to register the cpuidle driver for the 5440 as it has only
one WFI state which is the default idle function when the cpuidle driver is
disabled.

By disabling cpuidle we prevent to enter to the governor computation for
nothing, thus saving a lot of processing time.

The only drawback is the statistic via sysfs on this state which is lost but
it is meaningless and it could be retrieved from the ftrace easily.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Acked-by: Amit Kucheria <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Encapsulate boot vector code into a function for cpuidle

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Pass wakeup mask parameter to function for cpuidle

Pass the wakeup mask to 'exynos_set_wakeupmask' as this function could
be used for different idle states with different mask.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Remove ifdef for scu_enable in pm

The scu_enable function is already a noop in the scu's header file is
CONFIG_SMP=n, so no need to use these macros in the code.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move scu_enable in the cpu_pm notifier

We make the cpuidle code less arch dependent.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Use the cpu_pm notifier for pm

Use the cpu_pm_enter/exit notifier to group some pm code inside the
pm file.

The save and restore code is duplicated across pm.c and cpuidle.c. By
using the cpu_pm notifier, we can factor out the routine.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Fix S5P_WAKEUP_STAT call for cpuidle

This function should be called only when the powerdown sequence fails.

Even if the current code does not hurt, by moving this line, we have
the same code than the one in pm.c.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move some code inside the idle_finisher for cpuidle

Move the code around to differentiate different section of code and
prepare it to be factored out in the next patches.

The call order changed but hat doesn't have a side effect because
they are independent. The important call is cpu_do_idle() which must
be done the last.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Encapsulate register access inside a function for pm

That makes the code cleaner and encapsulted. The function will be
reused in the next patch.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Change function name prefix for cpuidle

The driver was initially written for exynos4 but the driver is used
also for exynos5.

Change the function prefix name exynos4 -> exynos

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Use cpuidle_register

Use the cpuidle generic function 'cpuidle_register'. That saves us
from some extra lines of code and unneeded variables.

A side effect of this change is a bug fix where before the cpuidle
driver was registered for each_online_cpu and now it is for
each_possible_cpu.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Prevent forward declaration for cpuidle

Move the structure below the 'exynos4_enter_lowpower' function so no
more need of forward declaration.

Signed-off-by: Daniel Lezcano <[email protected]>
Reviewed-by: Viresh Kumar <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Move arm core power down clock to exynos5250 common clock

Now with common clock support added for exynos5250 it is necessary to
move this code to exynos5250 common clock driver as clock registers
should be handled there. This change is tested in exynos5250 based
arndale platform.

Cc: Abhilash Kesavan <[email protected]>
Cc: Thomas Abraham <[email protected]>
Acked-by: Kukjin Kim <[email protected]>
Reviewed-by: Bartlomiej Zolnierkiewicz <[email protected]>
Signed-off-by: Amit Daniel Kachhap <[email protected]>
[t.figa: Rebased onto current kernel sources.]
Signed-off-by: Tomasz Figa <[email protected]>
Signed-off-by: Daniel Lezcano <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'v3.16-next/clk-samsung' into v3.16-next/cpuidle-exynos

cpufreq: exynos: Fix the compile error

Commit 7da83a80 ("ARM: EXYNOS: Migrate Exynos specific macros from
plat to mach") which lands in samsung tree causes build breakage
for cpufreq-exynos like following:

drivers/cpufreq/exynos-cpufreq.c: In function 'exynos_cpufreq_probe':
drivers/cpufreq/exynos-cpufreq.c:166:2: error: implicit declaration of function 'soc_is_exynos4210'
[-Werror=implicit-function-declaration]
drivers/cpufreq/exynos-cpufreq.c:168:2: error: implicit declaration of function 'soc_is_exynos4212'
[-Werror=implicit-function-declaration]
drivers/cpufreq/exynos-cpufreq.c:168:2: error: implicit declaration of function 'soc_is_exynos4412'
[-Werror=implicit-function-declaration]
drivers/cpufreq/exynos-cpufreq.c:170:2: error: implicit declaration of function 'soc_is_exynos5250'
[-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors
make[2]: *** [drivers/cpufreq/exynos-cpufreq.o] Error 1
make[2]: *** Waiting for unfinished jobs....
drivers/cpufreq/exynos4x12-cpufreq.c: In function 'exynos4x12_set_clkdiv':
drivers/cpufreq/exynos4x12-cpufreq.c:118:2: error: implicit declaration of function 'soc_is_exynos4212'
[-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors
make[2]: *** [drivers/cpufreq/exynos4x12-cpufreq.o] Error 1
make[1]: *** [drivers/cpufreq] Error 2

This fixes above error with getting SoC information via
of_machine_is_compatible() instead of soc_is_exynosXXXX().

Suggested-by: Tomasz Figa <[email protected]>
Signed-off-by: Jonghwan Choi <[email protected]>
[[email protected]: fixed typo and modified as per Viresh's suggestion]
[[email protected]: Rafael agreed]
Signed-off-by: Kukjin Kim <[email protected]>

ARM: S3C24XX: move debug-macro.S into the common space

Move debug-macro.S from mach/include to include/debug where
all other common debug macros are.

Signed-off-by: Heiko Stuebner <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: S3C24XX: use generic DEBUG_UART_PHY/_VIRT in debug macro

This removes the need for mach/-headers in the debug macro.

Signed-off-by: Heiko Stuebner <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: S3C24XX: trim down debug uart handling

Using the lowlevel debug uart is a corner case - even more so in a
multiplatform environment. So it seems reasonable to simply let the
developer set the appropriate uart type for the debugged SoC.

Signed-off-by: Heiko Stuebner <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: compressed/head.S: remove s3c24xx special case

addruart from the generic debug macro is doing exactly the same using
the common lowlevel uart definition, so there is no cause for this
special casing for s3c24xx.

Signed-off-by: Heiko Stuebner <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Remove unnecessary inclusion of cpu.h

Exynos specific macros and declarations have been moved to
mach-exynos. Inclusion of plat/cpu.h is no more necessary.

Signed-off-by: Sachin Kamat <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Migrate Exynos specific macros from plat to mach

Move Exynos specific macros to mach-exynos from plat-samsung to avoid
unnecessary dependency on plat based header files.

Signed-off-by: Sachin Kamat <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Remove exynos_subsys registration

'exynos_subsys' has no users. Remove this code.

Signed-off-by: Sachin Kamat <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Remove duplicate lines in Makefile

Group all files compiled under common config option together.

Signed-off-by: Sachin Kamat <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: use v7_exit_coherency_flush macro for cache disabling

A common macro v7_exit_coherency_flush available which does the below
tasks in the seqeunce.
-clearing C bit
-clearing L1 cache
-exit SMP
-instruction and data synchronization

So removing the local functions which does the same thing and use the
macro instead.

Signed-off-by: Leela Krishna Amudala <[email protected]>
Acked-by: Nicolas Pitre <[email protected]>
[[email protected]: tested on exynos3250 based board]
Tested-by: Chanwoo Choi <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'v3.16-next/clk-s3c24xx-3' into v3.16-next/cleanup-samsung

Documentation: sysfs: add Broadcom GISB arbiter sysfs properties

This patch documents the sysfs properties exposed by the Broadcom GISB
bus arbiter hardware.

Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: Arnd Bergmann <[email protected]>

Documentation: devicetree: add Broadcom GISB arbiter bindings

This patch adds the Broadcom GISB arbiter Device Tree binding that is
used on all Broadcom Set-top-box System-on-a-chip.

Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: Arnd Bergmann <[email protected]>

bus: add Broadcom GISB bus arbiter timeout/error handler

This patch adds support for the Broadcom GISB arbiter bus timeout/error
handler. GISB is a proprietary bus used by Broadcom Set Top Box
System-on-a-chip devices (BCM7xxx) which allows multiple masters and
clients to be interfaced with each other.

The bus arbiter offers support for generating two interrupts towards the
host CPU, thus allowing us to "catch" clock gated masters, or masters
being volontarily blocked for powersaving purposes, or do general system
troubleshooting.

We also register a hook with the ARM fault exception handling to allow
printing a more informative message than "imprecise external abort at
0x00000000" for instance.

Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: Arnd Bergmann <[email protected]>

Merge tag 'sunxi-drivers-for-3.16-2' of https://github.com/mripard/linux into next/drivers

Allwinner drivers changes for 3.16, take 2

Add reset driver for the A31

* tag 'sunxi-drivers-for-3.16-2' of https://github.com/mripard/linux:
  power: reset: Add Allwinner A31 reset code

Signed-off-by: Arnd Bergmann <[email protected]>

Merge tag 'qcom-drivers-for-3.16' of git://git./linux/kernel/git/galak/linux-qcom into next/drivers

Qualcomm ARM Based driver Updates for v3.16

* Introduce drivers/soc directory for misc SoC specific code
* Add driver to configure GSBI device

* tag 'qcom-drivers-for-3.16' of git://git.kernel.org/pub/scm/linux/kernel/git/galak/linux-qcom:
  soc: qcom: Add GSBI driver
  soc: Introduce drivers/soc place-holder for SOC specific drivers

Signed-off-by: Arnd Bergmann <[email protected]>

soc: qcom: Add GSBI driver

The GSBI (General Serial Bus Interface) driver controls the overarching
configuration of the shared serial bus infrastructure on APQ8064, IPQ8064, and
earlier QCOM processors.  The GSBI supports UART, I2C, SPI, and UIM
functionality in various combinations.

Signed-off-by: Andy Gross <[email protected]>
Signed-off-by: Kumar Gala <[email protected]>

soc: Introduce drivers/soc place-holder for SOC specific drivers

Based on earlier thread "https://lkml.org/lkml/2013/10/7/662" and
discussion at Kernel Summit'2013, it was agreed to create
'driver/soc' for drivers which are quite SOC specific.

Further discussion on the subject is in response to
the earlier version of the patch is here:
http://lwn.net/Articles/588942/

Cc: Greg Kroah-Hartman <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Paul Walmsley <[email protected]>
Cc: Olof Johansson <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Signed-off-by: Sandeep Nair <[email protected]>
Signed-off-by: Santosh Shilimkar <[email protected]>
Signed-off-by: Kumar Gala <[email protected]>

power: reset: Add Allwinner A31 reset code

That code used to be in the machine code, but it's more fit here with other
restart hooks.

That will allow to cleanup the machine directory, while waiting for a proper
watchdog driver for the A31.

Signed-off-by: Maxime Ripard <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>

ARM: edma: Remove redundant/unused parameters from edma_soc_info

The following parameters are no longer needed by the edma driver since the
information can be obtained from the IP's CCCFG register:
n_channel, n_region, n_slot and n_tc.
Remove the n_cc as well since in this context it has no meaning. We have
separate edma_soc_info struct/eDMA3_CC instance so this member does not
make any sense (and the driver no longer uses it).

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: davinci: Remove redundant/unused parameters for edma

The following parameters are no longer needed by the edma driver since the
information can be obtained from the IP's CCCFG register:
n_channel, n_region, n_slot and n_tc.
Remove the initialization of n_cc as well since in this context it has no
meaning. We have separate edma_soc_info struct/eDMA3_CC instance so this
member does not make any sense (and the driver no longer uses it).

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: dts: am4372: Remove obsolete properties from edma node

dma-channels, ti,edma-regions and ti,edma-slots no longer needed in DT since
the the same information is available in the IP's CCCFG register.

Signed-off-by: Peter Ujfalusi <[email protected]>
Acked-by: Tony Lindgren <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: dts: am33xx: Remove obsolete properties from edma node

dma-channels, ti,edma-regions and ti,edma-slots no longer needed in DT since
the the same information is available in the IP's CCCFG register.

Signed-off-by: Peter Ujfalusi <[email protected]>
Acked-by: Tony Lindgren <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

dt/bindings: ti,edma: Remove redundant properties from documentation

From CCCFG register of eDMA3 we can get all the needed information for the
driver about the IP:
Number of channels: NUM_DMACH
Number of regions: NUM_REGN
Number of slots (PaRAM sets): NUM_PAENTRY
Number of TC/EQ: NUM_EVQUE

The ti,edma-regions; ti,edma-slots and dma-channels in DT are
redundant since the very same information can be obtained from the HW.
The mentioned properties are deprecated.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Get IP configuration from HW (number of channels, tc, etc)

From CCCFG register of eDMA3 we can get all the needed information for the
driver about the IP:
Number of channels: NUM_DMACH
Number of regions: NUM_REGN
Number of slots (PaRAM sets): NUM_PAENTRY
Number of TC/EQ: NUM_EVQUE

In case when booted with DT or the queue_priority_mapping is not provided
set up a default priority map.

Signed-off-by: Peter Ujfalusi <[email protected]>
Acked-by: Joel Fernandes <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Save number of regions from pdata to struct edma

To be consistent in the code that we take parameters from edma_cc[j] struct
and not randomly from info[j] as well.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Remove num_cc member from struct edma

The struct edma is allocated per CC bases so the member num_cc does not make
any sense. One CC is one CC, it does not have sub CCs.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Remove queue_tc_mapping data from edma_soc_info

It is no longer in use by the driver or board files.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

Merge tag 'samsung-drivers' of git./linux/kernel/git/kgene/linux-samsung into next/drivers

Merge "Samsung driver update for 3.16" from Kukjin Kim:

exynos_mct update for v3.16

- register sched_clock callback to use clocksource mct-frc

* tag 'samsung-drivers' of http://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
  clocksource: exynos_mct: register sched_clock callback

Signed-off-by: Olof Johansson <[email protected]>

ARM: davinci: Remove eDMA3 queue_tc_mapping data from edma_soc_info

It is ignored by the edma driver since we are just setting back the default
mapping of TC -> Queue.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Do not change TC -> Queue mapping, leave it to default.

There is no need to change the default TC -> Queue mapping. By default the
mapping is: TC0 -> Q0, TC1 -> Q1, etc.
Changing this has no benefits at all and all the board files are just setting
the same mapping back to the HW.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Take the number of tc from edma_soc_info (pdata)

Instead of saving the for loop length, take the num_tc value from the pdata.
In case of DT boot set the n_tc to 3 as it is hardwired in edma_of_parse_dt()
This is a temporary state since upcoming patch(es) will change how we are
dealing with these parameters.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: No need to clean the pdata in edma_of_parse_dt()

The pdata has been just allocated with devm_kzalloc() in
edma_setup_info_from_dt() and passed to this function.

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

ARM: edma: Clean up and simplify the code around irq request

Get the two interrupt line number at the same time by merging the two
instance of if(node){}else{} places.
replace the &pdev->dev with the already existing dev which makes it possible
to collapse lines with devm_request_irq()

Signed-off-by: Peter Ujfalusi <[email protected]>
Signed-off-by: Sekhar Nori <[email protected]>

Merge branch 'sunxi/drivers-for-3.16' of https://github.com/mripard/linux into next/drivers

Merge "Allwinner drivers changes for 3.16" from Maxime Ripard:

Reset drivers for the A31, that have been merged with reset maintainer's
Acked-by.

* 'sunxi/drivers-for-3.16' of https://github.com/mripard/linux:
  reset: sunxi: allow MFD subdevices probe
  reset: sunxi: document sunxi's reset controllers bindings

Signed-off-by: Olof Johansson <[email protected]>

Merge tag 'mvebu-drivers-3.16' of git://git.infradead.org/linux-mvebu into next/drivers

Merge "ARM: mvebu: driver changes for v3.16" from Jason Cooper:

mvebu driver changes for v3.16

 - mvebu-devbus
    - changes need to add support for the orion5x platform

* tag 'mvebu-drivers-3.16' of git://git.infradead.org/linux-mvebu:
  memory: mvebu-devbus: add a devbus, keep-config property
  memory: mvebu-devbus: add Orion5x support
  memory: mvebu-devbus: split functions
  memory: mvebu-devbus: use _SHIFT suffixes instead of _BIT
  memory: mvebu-devbus: use ARMADA_ prefix in defines
  ARM: orion5x: fix target ID for crypto SRAM window
  memory: mvebu-devbus: fix the conversion of the bus width

Signed-off-by: Olof Johansson <[email protected]>

clk: exynos5420: Add 5800 specific clocks

Exynos5800 clock structure is mostly similar to 5420 with only
a small delta changes. So the 5420 clock file is re-used for
5800 also. The common clocks for both are seggreagated and few
clocks which are different for both are separately initialized.

Signed-off-by: Alim Akhtar <[email protected]>
Signed-off-by: Arun Kumar K <[email protected]>
Acked-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

dt-bindings: add documentation for Exynos3250 clock controller

The Exynos3250 clocks are statically listed and registered using the
Samsung specific common clock helper functions. Both device tree based
clock lookup and clkdev based clock lookups are supported.

Signed-off-by: Chanwoo Choi <[email protected]>
Signed-off-by: Tomasz Figa <[email protected]>
Acked-by: Kyungmin Park <[email protected]>
Cc: Mike Turquette <[email protected]>
Cc: Kukjin Kim <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Pawel Moll <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Ian Campbell <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Randy Dunlap <[email protected]>
Acked-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'for_3.16/exynos3250' of git://git./linux/kernel/git/tfiga/samsung-clk into v3.16-next/clk-samsung

Pull Samsung clock exynos3250 from Tomasz Figa:

"This small pull request contains a patch adding clock driver for Exynos3250,
which depends on previous pull requests in this series."

Acked-by: Mike Turquette <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'for_3.16/clk_cleanup' of git://git./linux/kernel/git/tfiga/samsung-clk into v3.16-next/clk-samsung

Pull Samsung clock cleanup from Tomasz Figa:

"This pull requests contains minor clean-up related to Samsung clock
support, including:

1) move Kconfig entries of Samsung clock drivers to drivers/clk,

2) compile drivers/clk/samsung conditionally when COMMON_CLK_SAMSUNG is
selected,

3) remove obsolete Kconfig lines after moving s3c24xx to CCF."

Conflicts:
arch/arm/mach-s3c24xx/Kconfig

Acked-by: Mike Turquette <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'for_3.16/clk_fixes_non_critical' of git://git./linux/kernel/git/tfiga/samsung-clk into v3.16-next/clk-samsung

Pull Samsung clock non-critical fixes from Tomasz Figa:

"This pull requests contains a number of non-critical fixes for Samsung clock
framework and drivers, including:

1) a series of fixes for Exynos5420 to correct clock definitions and make the
driver closer to the documentation,

2) several missing clocks and clock IDs added to Exynos4, Exynos5250 and
Exynos5420 drivers,

3) fix for incorrect initialization of clock table with NULL,

4) compiler warning fix."

Acked-by: Mike Turquette <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

Merge branch 'for_3.16/exynos5260' of git://git./linux/kernel/git/tfiga/samsung-clk into v3.16-next/clk-samsung

Pull Samsung clock exynos5260 from Tomasz Figa:

"This pull request contains patches preparing Samsung Common Clock Framework
helpers to support Exynos5260 by adding support for multiple clock providers
and then adding clock driver for Exynos5260."

Acked-by: Mike Turquette <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add MCPM call-back functions

Add machine-dependent MCPM call-backs for Exynos5420. These are used
to power up/down the secondary CPUs during boot, shutdown, s2r and
switching.

Signed-off-by: Thomas Abraham <[email protected]>
Signed-off-by: Inderpal Singh <[email protected]>
Signed-off-by: Andrew Bresticker <[email protected]>
Signed-off-by: Abhilash Kesavan <[email protected]>
Reviewed-by: Nicolas Pitre <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: dts: add CCI node for exynos5420

Add device-tree bindings for the ARM CCI-400 on Exynos5420. There
are two slave interfaces: one for the A15 cluster and one for the
A7 cluster.

Signed-off-by: Andrew Bresticker <[email protected]>
Signed-off-by: Abhilash Kesavan <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add generic cluster power control functions

Add generic cluster power control functions for exynos based SoCS
for cluster power up/down and to know the cluster status.

Signed-off-by: Abhilash Kesavan <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: use generic exynos cpu power control functions

Use generic exynos cpu power control functions to power up/down
and to know the status of the cpu in platsmp and hotplug code.

Signed-off-by: Leela Krishna Amudala <[email protected]>
Signed-off-by: Abhilash Kesavan <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: Add generic cpu power control functions for exynos SoCs

Add generic cpu power control functions for exynos based
SoCS for cpu power up/down and to know the cpu status.

Signed-off-by: Leela Krishna Amudala <[email protected]>
Signed-off-by: Abhilash Kesavan <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

reset: sunxi: allow MFD subdevices probe

The current implementation uses sunxi_reset_init function for both early
init and platform device probe.

The sunxi_reset_init function uses DT to retrieve device resources, which
will be an issue if reset controllers are registered from an MFD device
that define resources from mfd_cell definition.

Moreover, we can make of devm functions when we're in the probe context.

Signed-off-by: Boris BREZILLON <[email protected]>
Acked-by: Maxime Ripard <[email protected]>
Signed-off-by: Maxime Ripard <[email protected]>
Acked-by: Philipp Zabel <[email protected]>

reset: sunxi: document sunxi's reset controllers bindings

Add DT bindings documentation for sunxi's reset controllers.

Signed-off-by: Boris BREZILLON <[email protected]>
Acked-by: Maxime Ripard <[email protected]>
Signed-off-by: Maxime Ripard <[email protected]>
Acked-by: Philipp Zabel <[email protected]>

ARM: S3C24XX: fix merge conflict

Missed some changes during re-sorting this branch.
So fixed it.

Signed-off-by: Kukjin Kim <[email protected]>

ARM: dts: Remove g2d_pd node for exynos5420

G2D power domain also controls the CMU block of G2D. Since
clock registers can be accessed anytime for viewing
clk_summary, it can cause a system crash if g2d power domain
is disabled.

Signed-off-by: Arun Kumar K <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: dts: Remove mau_pd node for exynos5420

MAU powerdomain provides clocks for Audio sub-system block.
This block comprises of the I2S audio controller, audio DMA
blocks and Audio sub-system clock registers.

Right now, there is no way to hook up power-domains with
clock providers.  During late boot when this power-domain
gets disabled, we get following external abort.

Unhandled fault: imprecise external abort (0x1406) at 0x00000000
Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000007

Signed-off-by: Tushar Behera <[email protected]>
Tested-by: Doug Anderson <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: exynos_defconfig: enable HS-I2C to fix for mmc partition mount

High speed I2C is used on Exynos5 based SoCs. Enable it.

The MMC partition for Root filesystem cannot be mounted
without this enabling HS-I2C and regulators on many boards
are connected HS-I2C bus so the regulators don't come by
default without this.

Actually, we are not able to get arndale-octa board to boot
and mount an MMC partition without this change.

Signed-off-by: Sachin Kamat <[email protected]>
Tested-by: Javi Merino <[email protected]>
[[email protected]: modified description]
Signed-off-by: Kukjin Kim <[email protected]>

ARM: dts: disable MDMA1 node for exynos5420

This change places MDMA1 in disabled node for Exynos5420.
If MDMA1 region is configured with secure mode, it makes
the boot failure with the following on smdk5420 board.
("Unhandled fault: imprecise external abort (0x1406) at 0x00000000")
Thus, arndale-octa board don't need to do the same thing anymore.

Signed-off-by: Seungwon Jeon <[email protected]>
Tested-by: Javi Merino <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

ARM: EXYNOS: fix the secondary CPU boot of exynos4212

This patch fixes the offset of CPU boot address and changes
the parameter of smc call for SMC_CMD_CPU1BOOT command on
exynos4212.

Signed-off-by: Kyungmin Park <[email protected]>
Signed-off-by: Chanwoo Choi <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>

devicetree: bindings: document SYSRAM binding

Add SRAM binding documentation.

Signed-off-by: Sachin Kamat <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
[[email protected]: changed name SRAM to SYSRAM]
Signed-off-by: Kukjin Kim <[email protected]>