Migrate Tegra clock support to drivers/clk/tegra, this involves
moving:
1. definition of tegra_cpu_car_ops to clk.c
2. definition of reset functions to clk-peripheral.c
3. change parent of cpu clock.
4. Remove legacy clock initialization.
5. Initialize clocks using DT.
6. Remove all instance of mach/clk.h
Signed-off-by: NPrashant Gaikwad <pgaikwad@nvidia.com>
[swarren: use to_clk_periph_gate().]
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The device tree binding models Tegra30 CAR (Clock And Reset)
as a single monolithic clock provider.
Signed-off-by: NPrashant Gaikwad <pgaikwad@nvidia.com>
[swarren: fixed typo in binding doc]
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The Tegra20 CAR (Clock And Reset) Controller controls most aspects of
most clocks within Tegra20. The device tree binding models this as a
single monolithic clock provider, which exports many clocks. This reduces
the number of nodes needed in device tree to represent these clocks.

This binding is only useful for Tegra20; the set of clocks that exists on
Tegra30 is sufficiently different to merit its own binding.
Signed-off-by: NStephen Warren <swarren@nvidia.com>
Acked-by: NSimon Glass <sjg@chromium.org>
[pgaikwad: Added mux clk ids and sorted CAR node]
Signed-off-by: NPrashant Gaikwad <pgaikwad@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

tegra_cpu_car_ops struct is going to be accessed from drivers/clk/tegra.
Move the tegra_cpu_car_ops to include/linux/clk/tegra.h.
Signed-off-by: NPrashant Gaikwad <pgaikwad@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

Add function to read chip id from APB MISC registers. This function
will also get called from clock driver to flush write operations on
apb bus.
Signed-off-by: NPrashant Gaikwad <pgaikwad@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The "sleep.S" file has many functions that be shared by different module
currently. Not just for CPU idle driver. Make it build as default now.
Reported-by: NRhyland Klein <rklein@nvidia.com>
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
[swarren: add sleep.o to separate line so each line only contains 1 file]
Signed-off-by: NStephen Warren <swarren@nvidia.com>

It would rather to use the API of time_to_jiffies than a constant number
of jiffies for the wait time of CPU power up.

Based on the work by:
Sang-Hun Lee <sanlee@nvidia.com>
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The reset handler code is used for either UP or SMP. To make Tegra device
can compile for UP. It needs to be moved to another file that is not SMP
only. This is because the reset handler also be needed by CPU idle
"powered-down" mode. So we also need to put the reset handler init function
in non-SMP only and init them always.

And currently the implementation of the reset handler to know which CPU is
OK to bring up was identital with "cpu_present_mask". But the
"cpu_present_mask" did not initialize yet when the reset handler init
function was moved to init early function. We use the "cpu_possible_mask"
to replace "cpu_present_mask". Then it can work on both UP and SMP case.
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
[swarren: dropped the move of v7_invalidate_l1() from one file to another,
to avoid conflicts with Pavel's cleanup of this function, adjust Makefile
so each line only contains 1 file.]
Signed-off-by: NStephen Warren <swarren@nvidia.com>

There are some redundant codes in the CPUINIT section that was caused by
some codes not be organized well in "headsmp.S". Currently all the codes
in "headsmp.S" were put into CPUINIT section. But actually it doesn't
need to be loacted in CPUINIT section. There is no fuction access them
in CPUINIT section and we will relocate them to IRAM.

These codes also caused some unnecessary functions that access these
codes been put into CPUINIT section too. This patch clean it up and put
them into normal text section.
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The tegra_cpu_die was be executed by the CPU itslf. So the clock gating
procedure won't be executed after the CPU hardware shutdown code. Moving
the clock gating procedure to tegra_cpu_kill that will be run by another
CPU after the CPU died.
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
Acked-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

Updating the cache maintenance order before CPU shutdown when doing CPU
hotplug.
The old order:
* clean L1 by flush_cache_all
* exit SMP
* CPU shutdown
Adapt to:
* disable L1 data cache by clear C bit
* clean L1 by v7_flush_dcache_louis
* exit SMP
* CPU shutdown

For CPU hotplug case, it's no need to do "flush_cache_all". And we should
disable L1 data cache before clean L1 data cache. Then leaving the SMP
coherency.
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
Acked-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

The power up sequence is different on the cold boot CPU and the CPU
that resumed from the hotplug. For the cold boot CPU, it was been power
gated as default. To power up the cold boot CPU, the power should be
un-gated by un toggling the power gate register manually.

For the CPU that resumed from the hotplug, after un-halted the CPU. The
flow controller will un-gate the power of the CPU. No need to manually
control, just wait the power be resumed and continue the power up
sequence after the CPU power is ready.

Based on the work by:
Varun Wadekar <vwadekar@nvidia.com>
Signed-off-by: NJoseph Lo <josephl@nvidia.com>
Acked-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

tegra_cpu_init/exit will be called every time one cpu core is online or
offline. And all cpu cores share same clocks, redundant clk_get/put
wast time,  so I move them out.
Signed-off-by: NRichard Zhao <linuxzsc@gmail.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

Fix:

warning: (ARCH_TEGRA_2x_SOC) selects ARM_ERRATA_754327 which has unmet direct dependencies (CPU_V7 && SMP)
warning: (ARCH_TEGRA_2x_SOC) selects ARM_ERRATA_742230 which has unmet direct dependencies (CPU_V7 && SMP)

by selecting options only if SMP.
Signed-off-by: NStephen Warren <swarren@nvidia.com>

No need to be public. Checked with:
  $ touch arch/arm/mach-tegra/*[ch] && make C=1
Signed-off-by: NHiroshi Doyu <hdoyu@nvidia.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>

Move arch/arm/mach-tegra/timer.c to drivers/clocksource/tegra20_timer.c
so that the code is co-located with other clocksource drivers, and to
reduce the size of the mach-tegra directory.
Signed-off-by: NStephen Warren <swarren@nvidia.com>

After "ARM: delete struct sys_timer" board-rm9200-dt
fails compilation with the following error:

  CC      arch/arm/mach-at91/board-rm9200-dt.o
arch/arm/mach-at91/board-rm9200-dt.c:50:2: error: unknown field 'timer' specified in initializer
arch/arm/mach-at91/board-rm9200-dt.c:50:13: error: 'at91rm9200_timer' undeclared here (not in a function)
make[1]: *** [arch/arm/mach-at91/board-rm9200-dt.o] Error 1
make: *** [arch/arm/mach-at91] Error 2

This is a fall out from the timer conversion. Fix it by
converting board-rm9200-dt to use new timer init
function as well.
Signed-off-by: NJoachim Eastwood <manabian@gmail.com>
Signed-off-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

Commit 6bb27d73 "ARM: delete struct sys_timer" removed struct sys_timer,
but didn't update mach-sunxi/sunxi.c for this change, even though the
sunxi timer implementation itself was updated. This caused a build break:

arch/arm/mach-sunxi/sunxi.c:94:2: error: unknown field 'timer' specified in initializer
arch/arm/mach-sunxi/sunxi.c:94:12: error: 'sunxi_timer' undeclared here (not in a function)
Signed-off-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

I mismerged a previous branch from Alexander, and accidentally left
in ARCH_USES_GETTIMEOFFSET. Remove it.
Signed-off-by: NOlof Johansson <olof@lixom.net>
Cc: Alexander Shiyan <shc_work@mail.ru>

This fixes suspend to RAM adding necessary save and restore of L2 and GIC.
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

When we fail to power down, we need to clear out the power request.
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

With commit 384a2902 (ARM: gic: use a private mapping for CPU target
interfaces), wake-up IPIs now go to all cores as the gic cpu interface
numbering may not follow core numbering. This broke secondary boot on
highbank since the boot address was already set for all secondary cores,
this caused all cores to boot before the kernel was ready.

Fix this by moving the setting of the jump address to
highbank_boot_secondary instead of highbank_smp_prepare_cpus and
highbank_cpu_die. Also, clear the address when we boot. This prevents
cores from booting before they are actually triggered and is also necessary
to get suspend/resume to work.
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Signed-off-by: NOlof Johansson <olof@lixom.net>

s/hignbank/highbank/
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

With the addition of commit a0ae0240 (ARM: kernel: add device tree init
map function), the cpu reg values must match the cpu mpidr register or we'll
get warnings. For some reason, the CLUSTERID on highbank is 9, so the reg
value needs to be 0x90n to quiet the warnings.
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

While device_type is considered deprecated, it is still needed for tools
like lshw to identify cpu nodes.
Signed-off-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NOlof Johansson <olof@lixom.net>

The second FlexCAN port uses different clock than the first one, configure
correct clock to prevent hanging of the system during bringing up of the port.
Signed-off-by: NMarek Vasut <marex@denx.de>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: NSascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>

The IRQ array must be terminated by -1 and not by -1+OMAP_INTC_START
This led to having a resource list of 100s of IRQs.

Looks like this was caused by commit a2cfc509 (ARM: OMAP3+: hwmod: Add
AM33XX HWMOD data) that probably had some search and replace updates
done for the patch for sparse irq support.
Signed-off-by: NPantelis Antoniou <panto@antoniou-consulting.com>
Acked-by: NPaul Walmsley <paul@pwsan.com>
[tony@atomide.com: updated wit information about the breaking commit]
Signed-off-by: NTony Lindgren <tony@atomide.com>

During merge of the mvebu patches a clock gate for pinctrl was
lost. This patch just readds the clock gate.
Signed-off-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

The use of writel instead of writel_relaxed lead to deadlock in some
situation (SMP on Armada 370 for instance). The use of writel_relaxed
as it was done in the rest of this driver fixes this bug.
Signed-off-by: NGregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: NThomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: NJason Cooper <jason@lakedaemon.net>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This patch fixes a bug for Aurora L2 cache controller when the
write-through mode is enable. For the clean operation even if we don't
have to flush the lines we still need to invalidate them.
Signed-off-by: NGregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: NThomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: NJason Cooper <jason@lakedaemon.net>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

If CONFIG_ARCH_MULTIPLATFORM & CONFIG_ARCH_MVEBU are both enabled,
__v7_pj4b_setup is added between __v7_ca9mp_setup and __v7_setup.

But there's no jump instruction added. If the chip is Cortex A5/A9,
it goes through __v7_pj4b_setup also. It results in system hang.
Signed-off-by: NHaojian Zhuang <haojian.zhuang@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Since commit 62e4d357 (ARM: 7609/1: disable errata work-arounds which access
secure registers) ARM_ERRATA_743622/751472 depends on !ARCH_MULTIPLATFORM.

Since imx has been converted to multiplatform, the following warning happens:

$ make imx_v6_v7_defconfig
warning: (SOC_IMX6Q && ARCH_TEGRA_2x_SOC && ARCH_TEGRA_3x_SOC) selects
ARM_ERRATA_751472 which has unmet direct dependencies (CPU_V7 &&
!ARCH_MULTIPLATFORM)
warning: (SOC_IMX6Q && ARCH_TEGRA_3x_SOC) selects ARM_ERRATA_743622
which has unmet direct dependencies (CPU_V7 && !ARCH_MULTIPLATFORM)
warning: (SOC_IMX6Q && ARCH_TEGRA_3x_SOC) selects ARM_ERRATA_743622
which has unmet direct dependencies (CPU_V7 && !ARCH_MULTIPLATFORM)
warning: (SOC_IMX6Q && ARCH_TEGRA_2x_SOC && ARCH_TEGRA_3x_SOC) selects
ARM_ERRATA_751472 which has unmet direct dependencies (CPU_V7 &&
!ARCH_MULTIPLATFORM)

Recommended approach is to remove ARM_ERRATA_743622/751472 from being selected
by SOC_IMX6Q and apply such workarounds into the bootloader.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Acked-by: NRob Herring <rob.herring@calxeda.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The Armada XP MV78230 DT include file is missing a ; at the
end of the cpu node.
Reported-by: NThomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

We originally thought that the MV78230 variant of the Armada XP had
four Ethernet interfaces, like the other variants MV78260 and
MV78460. In fact, this is not true, and the MV78230 has only three
Ethernet interfaces.

So, the definitions of the Ethernet interfaces is now done as follows:

 * armada-370-xp.dtsi: definitions of the first two interfaces, that
   are common to Armada 370 and Armada XP

 * armada-xp.dtsi: definition of the third interface, common to all
   Armada XP variants.

 * armada-xp-mv78260.dtsi and armada-xp-mv78460.dtsi: definition of
   the fourth interface.
Signed-off-by: NThomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

Contrary to our understanding at the time armada-xp-mv78230.dtsi was
written, the MV78230 variant of the Armada XP SoC has two cores and
not one. This patch updates the .dtsi file to take into account this
reality.
Signed-off-by: NThomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

With the change to a DT based pinctrl/gpio driver, using gpio API
calls in board-*.c files no longer works, a dereferenced NULL pointer
exception occurs instead. By converting the GPIO code into a
fixed-regulator which gets probed later once pinctrl/gpio is
available, we avoid the exception.
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Tested-by: NStefan Peter <s.peter@mplch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

We moved to declaring clk gates in DT. However, device which do not
yet have a DT binding need to have a clkdev alias. This was missing
for SDIO.
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Tested-by: NStefan Peter <s.peter@mplch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

Clock Management of kirkwood has moved to DT clock providers.
However, TWSI1 has not yet been done.
This switches TWSI1 of 88f6282 to DT clock providers.
Signed-off-by: NNobuhiro Iwamatsu <iwamatsu@nigauri.org>
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

Without the clock being held by a driver, it gets turned off at a bad
time causing the SoC to lockup. This is often during reboot.
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Tested-by: NStefan Peter <s.peter@mpl.ch>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>

The UART controller used in the Armada 370 and Armada XP SoCs is the
Synopsys DesignWare 8250 (aka Synopsys DesignWare ABP UART). The
improper use of the ns16550 can lead to a kernel oops during boot if
a character is sent to the UART before the initialization of the
driver. The DW APB has an extra interrupt that gets raised when
writing to the LCR when busy. This explains why we need to use
dw-apb-uart driver to handle this.
Signed-off-by: NGregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: NJason Cooper <jason@lakedaemon.net>