The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Acked-by: NStephen Warren <swarren@nvidia.com>
Tested-by: NStephen Warren <swarren@nvidia.com>
Acked-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  We can remove the .init_machine as it becomes
the same as the generic version.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  This also allows us to eliminate the
.init_machine function as this becomes the same as the generic version.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add better commentry about the L2 cache requirements on these platforms.
Unfortunately, the auxiliary control register is not pre-set to indicate
the correct cache parameters, so we have to manually program these.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  Along with this change, we can delete l2x0.c
from prima2.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add support for L2 cache controller (PL310) on AM437x SoC.
Signed-off-by: NSekhar Nori <nsekhar@ti.com>
Acked-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Get rid of init call to initialize L2 cache.  Instead use the init_early
machine hook. This helps in using the initialization routine across
SoCs without the need of ugly cpu_is_*() checks.
Signed-off-by: NSekhar Nori <nsekhar@ti.com>
Acked-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

L2 cache initialization for OMAP4 redundantly sets the cache policy to
Round-Robin. This is not needed since thats the PL310 default anyway.

Removing this reduces the number of platform specific aux control
settings.
Signed-off-by: NSekhar Nori <nsekhar@ti.com>
Acked-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Avoid reading directly from the L2 registers in platform code.  The L2
code will have already saved the register values itself into the
l2x0_saved_regs structure, so platform code should just move these
values to where they're required.

This is safe because the L2x0 will have been initialised by an early
initcall, whereas the OMAP4 PM code is initialised late.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Since we now always enable NS access to the unlock registers, this can
be removed from OMAP4.  Remove the NS access bit for the interrupt
registers from OMAP4 as well - nothing in the kernel accesses that yet,
and we can add it in core code when we have the need.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Now that OMAP2 uses the write_sec method, we don't need to enable the L2
cache in OMAP2 specific code; this can be done via the normal mechanisms
in the L2C code.  Remove the OMAP2 specific code.
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

With the write_sec method, we no longer need to override the default
L2C disable method, and we no longer need the L2C set_debug method.
Both of these can be handled via the write_sec method.
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  This also allows us to eliminate the
.init_machine function as it is identical to the generic version.
Acked-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Acked-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.
Acked-by: NJason Cooper <jason@lakedaemon.net>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  Since the .init_irq method only calls
irqchip_init(), we can remove that too as the generic code will take
care of that.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Now that we handle this in core code, we don't need platforms enabling
the low power modes directly.
Acked-by: NShawn Guo <shawn.guo@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Now that highbank uses the write_sec method, we don't need to enable
the L2 cache in SoC specific code; this can be done via the normal
mechanisms in the L2C code.

Checking with Rob Herring:
> > Can we kill the "highbank_smc1(0x102, 0x1);" here?	That means
> > l2x0_of_init() will see the L2 cache disabled, and will try to enable
> > it via the write_sec hook, so it should do the right thing.
>
> Yes, that should work. You should be able to just call l2x0_of_init
> unconditionally. The condition was really to just avoid the smc on
> Midway which does get handled on h/w, but not if running virtualized.

So also drop the DT check too.  I'm leaving the config check in place
so that if L2 is disabled, the write_sec hook can be optimised away.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

With the write_sec method, we no longer need to override the default L2C
disable method.  This can be handled via the write_sec method instead.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

exynos was unconditionally calling the L2 cache initialisation from an
early_initcall.  This breaks multiplatform kernels.  Thankfully,
converting to generic l2c initialisation fixes this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The cache size should already be present in the L2 cache auxiliary
control register: it is part of the integration process to configure
the hardware IP.  Most platforms get this right, yet still many
cargo-cult program, and assume that they always need specifying to
the L2 cache code.  Remove them so we can find out which really need
this.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  We can remove the explicit machine init too
as this becomes identical to the generic version.
Acked-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove the explicit call to l2x0_of_init(), converting to the generic
infrastructure instead.  We can remove the explicit machine init too
as this becomes identical to the generic version.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Provide a common assembly implementation for PL310 resume code.  Certain
platforms need to re-initialise the L2C cache early as it may preserve
data across a S2RAM cycle, and therefore must be enabled along with the
L1 cache and MMU.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add a hook into the core ARM code to perform L2 cache initialisation
in a platform independent manner.  Platforms still get to indicate
their auxiliary control register values and mask, but the
initialisation call will now be made from generic code.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Since we always write to these during the cache initialisation, it is
a good idea to always have the non-secure access bit set.  Set it in
core code.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Always enable the L2C low power modes on L2C-310 R3P0 and newer parts.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Since we now automatically enable early BRESP in core L2C-310 code when
we detect a Cortex-A9, we don't need platforms/SoCs to set this bit
explicitly.  Instead, they should seek to preserve the value of bit 30
in the auxiliary control register.
Acked-by: NTony Lindgren <tony@atomide.com>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The AXI bus protocol requires that a write response should only be
sent back to the master when the last write has been accepted.  Early
BRESP allows the L2C-310 to send the write response as soon as the
store buffer accepts the write address.

Cortex-A9 processors can signal to the L2C-310 that they wish to be
notified early, and if this optimisation is enabled, the L2C-310 can
signal an early write response.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Move the L2 cache register saving to a more sensible location - after
the cache has been enabled, and fixups have been run.  We move the
saving of the auxiliary control register into the ->save function as
well which makes everything operate in a sane and maintainable way.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This is a required property, and should always be specified.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>