We no longer need the *_SEL* clock flags with the struct clksel conversion,
so we get rid of them.  This frees up 10 clock flag bits.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

RATE_CKCTL is no longer meaningful in the OMAP2 clock framework, so, drop it
from all clocks.  We keep the flag around since it's still used by OMAP1.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Simplify omap2_clk_set_rate() by separating out the clksel set_rate code into
its own function, omap2_clksel_set_rate().  Clksel clocks use this function
by placing its function pointer in their .set_rate fields.  Convert
existing clksel clocks accordingly.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

The OMAP clock framework struct clk contains a field, .round_rate,
that is intended to contain a function pointer to rate rounding code.
Take advantage of it by converting all clksel clocks' round_rate field
to point to omap2_clksel_round_rate().  Rename the existing
omap2_clksel_round_rate() to omap2_clksel_round_rate_div(), to signify
that it returns divisor information as well, and set up
omap2_clksel_round_rate() as a wrapper for the clock framework to use.
Simplify omap2_clk_round_rate() accordingly.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

src_offset and rate_offset are no longer used in OMAP2 clock framework
after the struct clksel/struct clksel_rate additions.  Remove these
fields from clock.h and #ifdef out those fields for OMAP24XX compiles.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

This completes the transition to struct clksel/clksel_rate by converting the
remaining clksel clocks to call omap2_clksel_recalc() on rate recalculation.
Also remove RATE_FIXED from some clocks that can now recalculate their own
rates correctly via clksel.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Get rid of now-obsolete code in omap2_clksel_recalc().
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Stop using clk->src_offset in omap2_clk_set_parent(); use what we get
from the clksel code instead.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Stop using clk->rate_offset in omap2_clk_set_rate(); use what we get
from the clksel code instead.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Convert omap2_clksel_get_src_field() to use new struct clksel and struct
clksel_rate data.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Convert omap2_get_clksel to use new struct clksel and struct
clksel_rate data.  Also fix its parameter order to conform with the
rest of the clock framework functions (i.e., struct clk first).
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Convert omap2_clksel_round_rate() to use the new clksel struct data.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Convert omap2_clksel_to_divisor() and omap2_divisor_to_clksel() to use
the new struct clksel and struct clksel_rate data.  Add
omap2_get_clksel_by_parent() utility code, called by both functions,
to return the appropriate struct clksel for a given (struct clk, parent
struct clk) combination.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Source-selectable clksel clocks have a 'default parent' assigned to them
at compile-time.  This default parent may or may not match the reality
that is configured in the hardware registers by the bootloader.  Clock tree
recalculations could be erratic if the struct clk parent field contents
don't match the hardware registers.

Resolve this by creating omap2_init_clksel_parent() to read the hardware
registers and update the struct clk parent field as appropriate for clksel
clocks.  Add an '.init' field to each source-selectable clk structure so
that the parent is fixed up when the clock is initially registered.  (We
don't do this for clksel clocks that are only rate-selectable, since they
only have one possible parent clock.)
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Many OMAP2 clocks are either source- or divisor-selectable by twiddling
various register bits; these are 'clksel' clocks.  The table of rate divisors
and source clock settings was previously embedded in the clock.c source code.
Encode this data in two data structures:

1) struct clksel_rate for divisor-selection, mapping divisor values to
register bit field values, along with a flags field indicating which
chips the rate is available for -- one for each parent-rate combination; and:

2) struct clksel for source-selection, mapping parent struct clk pointers
to struct clksel_rate array pointers; one for each parent.

Also, add several fields to the clk structure:

1) clksel: pointer to the struct clksel array

2) clksel_reg: pointer to the clksel register

3) clksel_mask: mask for the clksel bitfield

Add this data to each clksel clock.  All this data is not yet used by the clock
framework; that functionality is in the following patches.

N.B. Two clocks, sys_clkout and sys_clkout2, could not be converted to
this setup without splitting source selection and divisor selection.
This is because these clocks use separate registers for each
selector.  So, create two new clocks, sys_clkout_src and
sys_clkout2_src, and locate source-selection there, and keep divisor
selection in sys_clkout/sys_clkout2.  This entailed modifying
board-n800-audio.c.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

omap2_mpu_recalc() actually just looks up what the MPU speed 'should'
currently be from the current virt_prcm_set rate set, rather than reporting
the actual hardware settings.  Clarify this by renaming it to
omap2_table_mpu_recalc() and adding some comments.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

omap2_propagate_rate() seems completely superfluous.  It handles two
cases.  The first case is for non-fixed rate clocks.  It assigns the
parent's rate to the clock, and continues propagating the rate
downwards.  This can be handled by adding a RATE_PROPAGATES clock flag
to the clock, if it wasn't there already, and by using
followparent_recalc() directly as the recalc function.  The second
case is for fixed rate clocks.  Here all we need to do is to call
propagate_rate() directly, so, use that for the recalc function.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

The onchip_clks array is only used during init, so, mark it __initdata.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Add a new clock framework function, recalculate_root_clocks(), that
recalculates all root clocks (i.e., clocks with no parent clock).
Call this function in omap2_clk_arch_init(), rather than manually
propagating specific clocks.  Add propagate_rate() recalcs to func_32k_ck
and osc_ck.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

dsp_ick's parent clock is dsp_fck, not core_ck.  2420 TRM ES2.1.1 2.2 Rev Q
pp. 4-9, 5-28.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

This patch is to fix OMAP 3430 boot/compilation for updated mach-types.
Signed-off-by: Syed Mohammed Khasim <x0khasim@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

index c3eff5b..7b65e19 100644

This patch is to fix OMAP 3430SDP machine ID
Signed-off-by: Syed Mohammed Khasim <x0khasim@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

[   10.523437] BUG: sleeping function called from invalid context at kernel/mut6
[   10.523437] in_atomic():0, irqs_disabled():128
[   10.523437] [<c002c168>] (dump_stack+0x0/0x14) from [<c005374c>] (__might_sl)
[   10.523437] [<c0053698>] (__might_sleep+0x0/0xd4) from [<c024fdf4>] (mutex_l)
[   10.523437]  r5 = C02F0DE8  r4 = C02F0DF0
[   10.523437] [<c024fdd4>] (mutex_lock+0x0/0x44) from [<c0041df4>] (clk_get+0x)
[   10.523437]  r4 = 00000000
[   10.523437] [<c0041da4>] (clk_get+0x0/0x128) from [<c0046520>] (omap_getspee)
[   10.523437]  r8 = 00000002  r7 = 00000000  r6 = C031DAF8  r5 = C0473980
[   10.523437]  r4 = 00000000
[   10.523437] [<c00464fc>] (omap_getspeed+0x0/0x5c) from [<c01b8518>] (cpufreq)
[   10.523437]  r5 = C0473980  r4 = 00000002
Signed-off-by: Hiroshi DOYU <Hiroshi.DOYU@nokia.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Allow board code to disable use of DMA by passing in dma_channel = -1.
Signed-off-by: Kevin Hilman <khilman@mvista.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Keymap fix for f-sample and p2-sample.
Signed-off-by: Vivek Kutal <vivek.kutal@celunite.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Get rid of the following warnings:
arch/arm/mach-omap2/pm.c: In function 'omap2_enter_full_retention':
arch/arm/mach-omap2/pm.c:509: warning: implicit declaration of function 'omap2_gpio_prepare_for_retention'
arch/arm/mach-omap2/pm.c:536: warning: implicit declaration of function 'omap2_gpio_resume_after_retention'
Signed-off-by: Felipe Balbi <felipe.lima@indt.org.br>
Signed-off-by: Tony Lindgren <tony@atomide.com>

This should be the last of the CamelCase removal for CPPI...
more local variable updates:

 - "txState" --> tx_ram
 - "bReqComplete" --> complete (or remove)
 - "bdPtr" --> bd
 - "chanNum" --> index
 - remove pointless/unused "numCompleted"
 - convert cppi_rx_scan() to use bool, return true/false

Also improve indentation on the monstrous TX side of cppi_completion();
TX paths still deserve a rewrite to better match the RX side.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>

Even more CamelCase removal from CPPI:
 - members of cppi_channel struct
 - parameters to cppi_channel_program()
 - "buffSz" local variables
 - "chNum" local variables
 - "regVal" local variables

Also
 - remove the FIXME about using container_of(), that's resolved now
 - stop using the now-pointless private_data fields
 - remove the redundant actualLen field (use dma_channel.actual_len)
 - remove some redundent local variables
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>

More CamelCase removal from the CPPI support.  In this case:

 - CPPI descriptor members;
 - Some members of the controller struct;
 - Rename "regBase" variables (as "tibase", vs Mentor registers)
 - Rename otgCh variables as cppi_ch;

This code still has a bunch of CaMeLcAsInG going on.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>

Get rid of the following warning:
drivers/usb/musb/tusb6010.c: In function 'tusb_otg_ints':
drivers/usb/musb/tusb6010.c:716: warning: unused variable 'devctl'
Signed-off-by: Felipe Balbi <felipe.lima@indt.org.br>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Merge branches 'master' and 'omap-upstream'

Conflicts:

	arch/arm/configs/omap_h2_1610_defconfig
	arch/arm/configs/omap_osk_5912_defconfig
	arch/arm/mach-omap1/board-h2.c
	arch/arm/mach-omap1/board-palmte.c
	arch/arm/mach-omap1/board-palmtt.c
	arch/arm/mach-omap1/board-palmz71.c
	arch/arm/mach-omap1/board-sx1.c
	arch/arm/mach-omap2/Kconfig
	arch/arm/mach-omap2/Makefile
	arch/arm/mach-omap2/board-2430sdp.c
	arch/arm/mach-omap2/board-apollon-keys.c
	arch/arm/mach-omap2/board-apollon.c
	arch/arm/mach-omap2/board-h4.c
	arch/arm/mach-omap2/board-n800-audio.c
	arch/arm/mach-omap2/board-n800-dsp.c
	arch/arm/mach-omap2/board-n800-usb.c
	arch/arm/mach-omap2/board-n800.c
	arch/arm/mach-omap2/devices.c
	arch/arm/mach-omap2/gpmc.c
	arch/arm/mach-omap2/id.c
	arch/arm/mach-omap2/io.c
	arch/arm/mach-omap2/irq.c
	arch/arm/mach-omap2/memory.c
	arch/arm/mach-omap2/mux.c
	arch/arm/mach-omap2/pm.c
	arch/arm/mach-omap2/sleep.S
	arch/arm/plat-omap/Makefile
	arch/arm/plat-omap/common.c
	arch/arm/plat-omap/timer32k.c
	include/asm-arm/arch-omap/board-2430sdp.h
	include/asm-arm/arch-omap/eac.h
	include/asm-arm/arch-omap/gpio.h
	include/asm-arm/arch-omap/hardware.h
	include/asm-arm/arch-omap/io.h
	include/asm-arm/arch-omap/menelaus.h
	include/asm-arm/arch-omap/mmc.h
	include/asm-arm/arch-omap/omap24xx.h
	include/asm-arm/arch-omap/onenand.h
	include/asm-arm/arch-omap/pm.h

Add apollon gpio keys using gpio-keys input
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Update board Apollon with correct file path.
Signed-off-by: Trilok Soni <soni.trilok@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

This patch merges board specific files from N800 tree.
Nokia has published the files at:

http://repository.maemo.org/pool/maemo3.0/free/source/
kernel-source-rx-34_2.6.18.orig.tar.gz
kernel-source-rx-34_2.6.18-osso29.diff.gz
Signed-off-by: Kai Svahn <kai.svahn@nokia.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

- Add etherent gpmc handling
- Remove unused mux setting
- Add MMC switch pin comments
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

This patch syncs H4 board init with linux-omap tree.
Signed-off-by: Tony Lindgren <tony@atomide.com>

Replace OMAP1610 with OMAP2420.
Signed-off-by: Komal Shah <komal_shah802003@yahoo.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

H4 has two peripheral ports, one for "download" and one for OTG.
The one to use is selected through Kconfig.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Tony Lindgren <tony@atomide.com>

The PRCM base register is different on 242x and 243x.  Use
the #ifdef'd #define from omap24xx.h instead of the locally
defined one.
Signed-off-by: Kevin Hilman <khilman@mvista.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>

Device init for OMAP24xx Enhanced Audio Controller
Signed-off-by: Jarkko Nikula <jarkko.nikula@nokia.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>