We can currently get "Unable to handle kernel paging request at
virtual address" for invalid clocks with dts node but no driver:

(__clk_get_hw) from [<c0138ebc>] (ti_sysc_find_one_clockdomain+0x18/0x34)
(ti_sysc_find_one_clockdomain) from [<c0138f0c>] (ti_sysc_clkdm_init+0x34/0xdc)
(ti_sysc_clkdm_init) from [<c0584660>] (sysc_probe+0xa50/0x10e8)
(sysc_probe) from [<c065c6ac>] (platform_drv_probe+0x58/0xa8)

Let's add IS_ERR checks to ti_sysc_clkdm_init() as And let's start treating
clk_get() with -ENOENT as a proper error. If the clock name is specified
in device tree we must succeed with clk_get() to continue. For modules with
no clock names specified in device tree we will just ignore the clocks.

Fixes: 2b2f7def ("bus: ti-sysc: Add support for missing clockdomain handling")
Acked-by: NRoger Quadros <rogerq@ti.com>
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

NFSroot can fail on dra7 when cpsw is probed using ti-sysc interconnect
target module driver as reported by Keerthy.

Device clocks and the interconnect target module may or may not be
enabled by the bootloader on init, but we currently assume the clocks
and module are on from the bootloader for "ti,no-idle" and
"ti,no-idle-on-init" quirks as reported by Grygorii Strashko.

Let's fix the issue by always enabling clocks init, and
never disable them for "ti,no-idle" quirk. For "ti,no-idle-on-init"
quirk, we must decrement the usage count later on to allow PM
runtime to idle the module if requested.

Fixes: 1a5cd7c2 ("bus: ti-sysc: Enable all clocks directly during init to read revision")
Cc: Keerthy <j-keerthy@ti.com>
Cc: Vignesh Raghavendra <vigneshr@ti.com>
Reported-by: NKeerthy <j-keerthy@ti.com>
Reported-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

The clocks are not yet parsed and prepared until after a successful
sysc_get_clocks(), so there is no need to unprepare the clocks upon
any failure of any of the prior functions in sysc_probe(). The current
code path would have been a no-op because of the clock validity checks
within sysc_unprepare(), but let's just simplify the cleanup path by
returning the error directly.

While at this, also fix the cleanup path for a sysc_init_resets()
failure which is executed after the clocks are prepared.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We are currently using a wrong register for dcan revision. Although
this is currently only used for detecting the dcan module, let's
fix it to avoid confusion.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We have cases where there are no softreset bits like with am335x lcdc.
In that case ti,sysc-mask = <0> needs to be handled properly.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

For some devices we can get the following warning on boot:

ti-sysc 48485200.target-module: sysc_disable_module: invalid midlemode

Fix this by treating SYSC_IDLE_FORCE like we do for the other bits
for idlemodes mask.

Fixes: d59b6056 ("bus: ti-sysc: Add generic enable/disable functions")
Cc: Roger Quadros <rogerq@ti.com>
Reviewed-by: NSuman Anna <s-anna@ti.com>
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Some older interconnect target modules need module internal clock
toggling quirks to reset properly. We've been doing this in the
platform code earlier, but need to be able to it directly in the
ti-sysc driver when we no longer rely on on the platform code.

Let's add reset handling for 1-wire, i2c and watchdog. Later on
we can add more modules like msdi and dss as they get tested.
For dra7 pcie, we should be able to just use the rstctrl reset
driver when available.
Signed-off-by: NTony Lindgren <tony@atomide.com>

Looks like we currently only detect UART on omap36xx, let's also
add support for omap34xx. And let's also fix the SWSUP mode, it should
be SWSUP_SIDLE for omap3, not SWSUP_SIDLE_ACT like for omap4 and later.

Note that we are still booting omap3 for most part without ti-sysc,
so no need to treat this change as a fix.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We need to deassert rstctrl resets before enabling clocks to avoid clock
"failed to enable" errors. For asserting rstctrl reset, the clocks need
to be enabled.

As the reset controller status is not available for arrays, let's use
devm_reset_control_get_optional() so we can get the status after reset.

Note that depends on a proper PRM rstctrl driver, so far I've only
tested this with earlier reset-simple patches.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We must not assert reset for modules with no child device drivers
until in runtime_suspend. Otherwise register access will fail without
legacy mode helping us.

Let's add a flag for disable_on_idle and move the reset driver
handling to runtime suspend and resume. We can then also use the
disable_on_idle flag to reconfigure sysconfig register for PM
modes requesting it.

Let's also make the other flags use bitfield while at it instead of
bool.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Some modules have ENAWAKEUP bit that we need to configure when not
relying on platform data callbacks.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Some modules have idlemodes wired, but not completely functional. We have
quirks for SWSUP_SIDLE and SWSUP_SIDLE_ACT to manage interconnect target
modules without hardware support, but we've been only using them so far
in legacy mode. Let's add support for SWSUP quirks in non-legacy mode too.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Modules with clockactivity need it configured during enable.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

For interconnect target modules with autoidle bit wired, we need to manage
it for enable and disable.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We need to specify QUIRK_LEGACY_IDLE for device drivers that still have
pm_runtime_irq_safe() set like 8250.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We've had minimal OCP softreset support in ti-sysc interconnect target
module driver only used for MCAN driver so far. But it turns out that
MCAN has the sysstatus register resetdone bit inverted compared to most
other modules.

Let's make OCP softreset work for other typical cases with reset status
in sysstatus or sysconfig register so we can use the new functions for
sysc_enable_module() and sysc_disable_module() without "ti,hwmods"
property in the following patches.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We need to also support 16-bit writes for i2c in addition to the reads
when we start configuring the sysconfig register for reset and idle modes.

Note that only i2c revision register has LO and HI registers, so let's
add a check also for 16-bit register read. This change is currently cosmetic
and does not affect anything until we enable the module specific quirk
handling for i2c reset and enable later on.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We need to let ti-sysc driver manage clockdomain autoidle for the
duration of of reset, enable and idle. And we need to do it before we
enable the clock and after we disable it. Currently we are still
relying on platform callbacks indirectly managing clockdomain autoidle.
But I noticed that for device tree only probed drivers it now happens
only after we enabling the clocks and before we disable the clocks,
while it should be the other way around. So far I have not noticed
any issues with this though.

Let's add new ti_sysc_clkdm_deny_idle() and ti_sysc_clkdm_allow_idle()
functions for ti-sysc driver to use to manage clockdomains directly via
platform data callbacks. Note that we can implement the clockdomain
functions in pdata-quirks.c as for probing devices without "ti,hwmods"
custom property we don't need to use the other platform data callbacks.

Let's do this in one patch as there's is still an unlikely chance we
may need to apply this as a fix for v5.2 for dropping legacy platform
data for some devices. We also do have the option of adding back the
platform data if needed in case of trouble.
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Some interconnect target modules have no module control registers at
all, such as d_can on am335x and am437x.

The d_can register offset at 0 is CTL register with 0x401 as the default
value. I guess I mistook the 0x401 value for a revision register as the
value happens to look similar to what the revision registers typically
have for other modules.

To handle modules with no control registers, we need to improve the
ti-sysc driver a bit to bail out with errors on no control registers,
and then we can remove the bogus revision registers for d_can.

Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Tested-by: NKeerthy <j-keerthy@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

For non legacy cases, add generic sysc_enable_module()
and sysc_disable_module() functions.
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Detect DMIC to see what we have connected if config DEBUG is enabled.
Signed-off-by: NTony Lindgren <tony@atomide.com>

In preparation of dropping interconnect target module platform data in
favor of devicetree based data, we must pass swsup idle quirks to the
platform data functions.

For now, let's only tag the UART modules with the SWSUP_SIDLE_ACT quirk.
The other modules will get tagged with swsup quirks as we drop the
platform data and test the changes.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We cannot access mcpdm registers at all unless there is an optional pdmclk
configured. As this is currently only needed for mcpdm, let's check for
mcpdm in sysc_get_clocks(). If it turns out to be needed for other modules
too, we can add more flags to the quirks table for this.
Signed-off-by: NTony Lindgren <tony@atomide.com>

At least mcpdm needs an optional external clock enabled to function and
this clock typically comes from the PMIC. We can detect mcpdm based on
the interconnect target module address and set a quirk flag early.

To do this, let's initialize the clocks a bit later and add a new
function for sysc_init_early_quirks(). Note that we cannot yet enable
the early quirks for mcpdm until the optional external clocks are
handled in the in the following patch.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We can do the rsstctrl a bit later, but need to deassert rstctrl reset
before the clocks are enabled if asserted. Let's only init restctrl
in sysc_init_resets() and do the reset later on just before we enable
the device clocks.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We are currently not managing interconnect target module clocks in the
for legacy platform data based case. This causes a problem for using the
platform data based functions when dropping the platform data for the
interconnect target module configuration.

To avoid a situation where we need to populate the main and optional
clocks also for the platform data based functions, let's just manage the
clocks directly in ti-sysc driver. This means that until the interconnect
target module confugration platform data is dropped our use count for
clk_enable() will be 2 instead of 1.
Signed-off-by: NTony Lindgren <tony@atomide.com>

The platform data based init functions typically reset the interconnect
target module configure the registers. As we may need the interconnect
target module specific quirks configured based on the revision register,
we want to move the platform data based init to happen later.

Let's allocate mdata as needed so it's available for sysc_legacy_init()
that we call with module clocks enabled from sysc_init_module().
Signed-off-by: NTony Lindgren <tony@atomide.com>

The first thing we want to do is just read the module revision register to
be able to configure the module specific quirks and configure the module
registers.

As the interconnect target module may not yet be properly configured and
may need a reset first, we don't want to use pm_runtime_get() at this
point.

To read the revision register, let's just enable the all the clocks for
the interconnect target module during init even if the optional clocks
are not needed. That way we can read the revision register to configure
the quirks needed for PM runtime.
Signed-off-by: NTony Lindgren <tony@atomide.com>

At least McPDM module depends on an external optional clock to be
usable. To make handling of the McPDM clock easier in the following
patches, let's add separate functions for handling the main clocks
and the optional clocks.

Let's also add error handling to shut down already enabled clocks
while at it.
Signed-off-by: NTony Lindgren <tony@atomide.com>

Let's move the legacy idle and enable into separate functions to simplify
PM runtime functions a bit.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We can make sysc_write() and sysc_child_pm_domain static as noted by
sparse.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We have ti,no-idle in use in addition to ti,no-idle-on-init but we're
missing handling for it in the ti-sysc interconnect target module driver.

Let's also group the idle defines together and update the binding
documentation for it.

Cc: devicetree@vger.kernel.org
Reviewed-by: NRob Herring <robh@kernel.org>
Signed-off-by: NTony Lindgren <tony@atomide.com>

If we return early before ddata->clocks have been allocated we will get a
NULL pointer dereference in sysc_unprepare(). Let's fix this by returning
early when no clocks are allocated.

Fixes: 0eecc636 ("bus: ti-sysc: Add minimal TI sysc interconnect target driver")
Signed-off-by: NTony Lindgren <tony@atomide.com>

Commit 84badc5e ("ARM: dts: omap4: Move l4 child devices to probe
them with ti-sysc") started producing a warning for pwm-omap-dmtimer:

WARNING: CPU: 0 PID: 77 at drivers/bus/omap_l3_noc.c:147
l3_interrupt_handler+0x2f8/0x388
44000000.ocp:L3 Custom Error: MASTER MPU TARGET L4PER2 (Idle):
Data Access in Supervisor mode during Functional access
...
__pm_runtime_idle
omap_dm_timer_disable
pwm_omap_dmtimer_start
pwm_omap_dmtimer_enable
pwm_apply_state
pwm_vibrator_start
pwm_vibrator_play_work

This is because the timer that pwm-omap-dmtimer is using is now being
probed with ti-sysc interconnect target module instead of omap_device
and the ti-sysc quirk for SYSC_QUIRK_LEGACY_IDLE is not fully
compatible with what omap_device has been doing.

We could fix this by reverting the timer changes and have the timer
probe again with omap_device. Or we could add more quirk handling to
ti-sysc driver. But as these options don't work nicely as longer term
solutions, let's just make timers probe with ti-sysc without any
quirks.

To do this, all we need to do is remove quirks for timers for ti-sysc,
and drop the bogus pm_runtime_irq_safe() flag for timer-ti-dm.

We should not use pm_runtime_irq_safe() anyways for drivers as it will
take a permanent use count on the parent device blocking the parent
devices from idling and has been forcing ti-sysc driver to use a
quirk flag.

Note that we will move the timer data to DEBUG section later on in
clean-up patches.

Fixes: 84badc5e ("ARM: dts: omap4: Move l4 child devices to probe them with ti-sysc")
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: H. Nikolaus Schaller <hns@goldelico.com>
Cc: Keerthy <j-keerthy@ti.com>
Cc: Ladislav Michl <ladis@linux-mips.org>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Sebastian Reichel <sre@kernel.org>
Cc: Tero Kristo <t-kristo@ti.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Reported-by: NH. Nikolaus Schaller <hns@goldelico.com>
Tested-By: NAndreas Kemnade <andreas@kemnade.info>
Tested-By: NH. Nikolaus Schaller <hns@goldelico.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

With ti-sysc, we need to now have the device tree properties for
ti,no-reset-on-init and ti,no-idle-on-init at the module level instead
of the child device level.

Let's check for these properties at the child device level to enable
quirks, and warn about moving the properties to the module level.

Otherwise am335x-evm based boards tagging gpio1 with ti,no-reset-on-init
will have their DDR power disabled if wired up in such a tricky way.

Note that this should not be an issue for earlier kernels as we don't
rely on this until the dts files have been updated to probe with ti-sysc
interconnect target driver.

Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reported-by: NPeter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We want to see the names of detected devices when DEBUG is enabled.
Signed-off-by: NTony Lindgren <tony@atomide.com>

We have OPT_CLKS_NEEDED in legacy platform data, but it's missing
from the ti-sysc driver for device tree based configuration.

In order to pass OPT_CLKS_NEEDED quirk flag we need to update omap4 module
data and add a new compatible for dra7 as the module layout is different
from sysc_regbits_omap4_mcasp.

Fixes: 70a65240 ("bus: ti-sysc: Add register bits for interconnect
target modules")
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Rob Herring <robh+dt@kernel.org>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We can have holes in clock_roles with interface clock missing for
example. Currently getting an optional clock will fail if there are
only a functional clock and an optional clock.

Fixes: 09dfe581 ("bus: ti-sysc: Add handling for clkctrl opt clocks")
Signed-off-by: NTony Lindgren <tony@atomide.com>

As Grygorii Strashko pointed out, the runtime PM use count of the
children can be whatever at suspend and we should not use it. So
let's just suspend ti-sysc at noirq level and get rid of some code.

Let's also remove the PM_SLEEP ifdef and use __maybe_unused as the
PM code already deals with the ifdefs.
Suggested-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

We're currently warning about busy children on suspend in
sysc_child_suspend_noirq() but the legacy code omap_device does
not do that. Let's just make it dev_dbg() instead of dev_warn().
Signed-off-by: NTony Lindgren <tony@atomide.com>