With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Since mxs is a DT only platform now, along with the changes, the non-DT
case handling in probe function also gets removed.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Since mxs is a DT only platform now, along with the changes, the non-DT
case checking in probe function also gets cleaned up.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Chris Ball <cjb@laptop.org>
Cc: linux-mmc@vger.kernel.org
Reviewed-by: Arnd Bergmann <arnd@arndb.de>

Update mxs-dma driver to adopt generic DMA device tree binding.  It
calls of_dma_controller_register() with mxs specific of_dma_xlate to
get the generic DMA device tree helper support.  Then DMA clients only
need to call dma_request_slave_channel() for requesting a DMA channel
from dmaengine.

The existing way of requesting channel, clients directly call
dma_request_channel(), still work there, and will be removed after
all mxs-dma clients get converted to generic DMA device tree helper.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>

Use devm_* managed functions to simplify probe() error handling.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>

Add generic DMA device tree binding for mxs-dma.  The changes include:

 * Add channel interrupts into DMA controller nodes
 * Add properties '#dma-cells' and 'dma-channels' for DMA controller nodes
 * And properties 'dmas' and 'dma-names' for DMA client nodes
 * Update mxs-dma device tree binding doc
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>

Some pins on the i.mx23 and i.mx28 are missing pull-ups, document that
oddity because it was difficult to know the expected behaviour.
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

None of mxsfb users uses mxsfb_platform_data now.  Let's remove it
from mxsfb driver.

As the result, include/linux/mxsfb.h gets deleted with a few macros
moved into mxsfb.c.  Along with the change, the typo "FAILING" in macro
name is fixed to be "FALLING".
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Move display timing configurations into device tree, so that the
auxdata for mxsfb driver can be killed.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Use videomode helpers to get display timings and configurations from
device tree when platform_data is absent.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

There is no in-tree mxsfb users using mxsfb_platform_data dotclk_delay.
Let's remove it from mxsfb_platform_data to ease full device tree
adoption of mxsfb driver.  If later we have platform/board need to
configure this parameter, we can add it into device tree bindings.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

There is no in-tree users of mxsfb_platform_data fb_phys/fb_size.
With CMA support in the kernel, there is no real need for platform to
reserve memory and pass address and size into driver via platform_data.
So let's remove fb_phys/fb_size from mxsfb_platform_data to ease full
device tree adoption.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Use devm_* managed functions to make code a little cleaner.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

With fec driver taking care of enet_out clk, most of board setup
do not need to enable enet_out clk.  So remove them.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Put the clock to the devicetree, so the driver can take care of it
later. Then, we don't have to do the enabling as a workaround in board
init.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
[shawn.guo: add enet_out into imx28.dtsi and overwrite it for m28evk]
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Some MX28 boards need the internal enet_out clock to be enabled. So, do
this in the driver iff the clock was referenced via devicetree.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: David S. Miller <davem@davemloft.net>

When the optional clk_ptp is absent, we can just set it to NULL, and
clk API will just handle it gracefully.  It saves us from checking
clk_ptp whenever calling into clk API.

Also since clk_ptp is optional, the "ret" variable shouldn't be set
in case that the clock is absent.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: David S. Miller <davem@davemloft.net>

Fix the following build error:

drivers/clocksource/mxs_timer.c:304:1: error: expected ',' or ';' at end of input
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Most of the stuff in the headers are used nowhere now.  Move a few
things that are useful for mach-mxs.c into there and remove the headers.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Most of the function hooks are same between imx23 and imx28
machine_desc, so merge them into one.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

All three remaining functions declared in common.h are implemented by
clock driver.  Create header include/linux/clk/mxs.h to contain them
and remove common.h.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

All the users of mxs_get_ocotp() are in mach-mxs.c.  Move the function
into mach-mxs.c, make it a static function, and then remove ocotp.c.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

The static mapping is used nowhere now.  Hence mm.c can be removed
completely.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

The only user of the static mapping done in mx23_map_io and mx28_map_io
is low-level debug now.  Use debug_ll_io_init() instead, so that the
static mapping is used nowhere and can be removed completely later.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Instead of using the static definitions, get ocotp base address from
device tree with mapping.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

There is no user of function mxs_reset_block() now.  Let's move
mxs_restart() into mach-mxs.c as a static function and remove system.c
completely.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Instead of using static address definition, get reset address from
device tree with mapping, so that core_initcall mxs_arch_reset_init()
can be killed.

The "rtc" clock code in mxs_arch_reset_init() seems to be zombie, since
there is no clk lookup defined in clock driver at all.  Remove it
together.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

The hardware.h is an empty header and used nowhere now.  Remmove it.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

The header <mach/hardware.h> is not needed at all, and <mach/mxs.h> is
needed only for macros MXS_SET_ADDR and MXS_CLR_ADDR.  Define the macros
and remove the mach header inclusions.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Headers <mach/mxs.h> and <mach/common.h> are not used in the driver
at all.  Removed the inclusions.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Jonathan Cameron <jic23@kernel.org>

The function stmp_reset_block() provides the exactly same functionality
as mxs_reset_block().  So use stmp_reset_block() instead, so that
<mach/common.h> inclusion can be removed.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: rtc-linux@googlegroups.com

It removes the use of mach level IO accessor __mxs_setl/clrl, and hence
removes mach header inclusion from clock driver.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mike Turquette <mturquette@linaro.org>

Instead of using the static definitions, get clkctrl and digctl base
addresses with mapping from device tree.

Use macro on variable is not nice, but it's done here to save huge
pointless diff stat.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mike Turquette <mturquette@linaro.org>

Remove the unneeded mach-types.h inclusion from mxs.h, which is a
leftover from commit 845da6bf (ARM: mxs: detect SoC by checking CHIPID
register).
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Move icoll.c into drivers/irqchip as irq-mxs.c, and along with the
renaming, change the driver to use IRQCHIP_DECLARE.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Call stmp_reset_block() rather than mxs_reset_block(), so that
<mach/common.h> inclusion can be removed from icoll driver.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Rather than using the static definition, it gets icoll base address
with mapping from device tree.

As the result, <mach/mxs.h> inclusion can be removed from the driver
now.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Move mxs timer driver into drivers/clocksource as mxs_timer.c.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Select STMP_DEVICE and in timer code replace mxs_reset_block() with
stmp_reset_block(), use STMP_OFFSET_REG_SET/CLR to replace
__mxs_setl/clrl.

As the result, <mach/mxs.h> and <mach/common.h> includsion can be
removed from timer.c now.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

Remove cpu_is_mx23() call from timer code by using
of_device_is_compatible() instead.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>