Add SoC specific map_io function to be used by the generic DT
board file. This is an intermediate step before having some
generic DT aware map_io function.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
This generalizes the omap2_mcbsp1_mux_clkr_src and omap2_mcbsp1_mux_fsr_src
implementation between generic McBSP and OMAP2 specific McBSP code. These
functions are used to select source for CLKR and FSR signals on OMAP2+.
Start generalizing the code by implementing an optional mux_signal function
pointer in platform data that will implement the actual muxing and which is
called now from omap2_mcbsp1_mux_clkr_src and omap2_mcbsp1_mux_fsr_src.
These functions are to be removed later and cleanup the API so that
mux_signal gets its arguments directly from client code.
Signed-off-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
This generalizes the omap2_mcbsp_set_clks_src implementation between generic
McBSP and OMAP2 specific McBSP code. Currently this function is used to
select either internal fclk or clks pin as a McBSP CLKS source on OMAP2+.
Implement generalization by having an optional set_clk_src function pointer
in platform data that is used to select parent for a given clock. Idea is to
pass higher level source clock name (later coming from client driver) that
platform specific code will map to platform specific clock name.
API cleanup between McBSP and client code comes later.
Signed-off-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Active sidetone requires that McBSP interface clock doesn't idle and there
is no mechanism in hwmod to turn autoidling on/off in runtime. McBSP2 and 3
in OMAP34xx share their interface clock with McBSP sidetone module and
that interface clock must be active when the sidetone is operating.
Sidetone has its own autoidle bit which should keep the interface clock
active but it is broken. Putting the McBSP core to no-idle mode when the
sidetone is active is no good either since it results to higher power
consumption when using the threshold based DMA transfers.
For making the McBSP code more generic, move this sidetone clock management
with fixme comments to mach-omap2/mcbsp.c and pass pointer to it via
platform data.
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Cc: Paul Wamsley <paul@pwsan.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
McBSP transmit and receive configuration control registers must be set up
for OMAP2430 and later. Replace is_omap tests in generic code with a new
feature flag has_ccr in platform data so that there is no need to change
code for any upcoming OMAP version.
Signed-off-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Currently wakeup control code is compiled only when CONFIG_ARCH_OMAP3 is
set even it should be available for CONFIG_ARCH_OMAP4 only builds also.
Fix this by making wakeup control generic so that it is executed whenever
new feature flag has_wakeup in platform data is set. Currently flag is set
for McBSP config types 3 and 4.
Remove also old comments about idle mode settings and HW bug workarounds
that were not updated during hwmod conversion.
Signed-off-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Register access can be made more generic by calculating register address
offsets runtime from common register definitions and by using reg_size and
reg_step variables that are passed via platform data. Common register
definitions are possible since McBSP registers are ordered similarly between
OMAP versions.
Remove also references to OMAP2+ specific config_type variable from generic
McBSP code since other variables and feature flags are better to carry needed
information from platform code.
Signed-off-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Commit f41caddbe7 (omap2+: Use
Kconfig symbol in Makefile instead of obj-y) cleaned up the
omap2+ Makefile. However this did not account for the inline
functions that are now needed for board_flash_init and
board_nand_init.
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Tony Lindgren <tony@atomide.com>
During the idle/suspend path, we expect the console lock to be held so
that no console output is done during/after the UARTs are idled.
However, when using the no_console_suspend argument on the
command-line, the console driver does not take the console lock. This
allows the possibility of console activity after UARTs have been
disabled.
To fix, update the current is_suspending() to also check the
console_suspend_enabled flag.
Reported-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
Tested-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Rather than embedding a struct platform_device inside a struct
omap_device, decouple them, leaving only a pointer to the
platform_device inside the omap_device.
Use the arch-specific data field of the platform_device (pdev_archdata)
to add an omap_device pointer after the platform_device has been created.
Signed-off-by: Kevin Hilman <khilman@ti.com>
Public omap_device functions need to take platform_device pointers,
conversion to omap_device pointers is done internal to the omap_device
layer.
Signed-off-by: Kevin Hilman <khilman@ti.com>
All of the device init and device driver interaction with omap_device
is done using platform_device pointers. To make this more explicit,
have omap_device return a platform_device pointer instead of an
omap_device pointer.
All current users of the omap_device pointer were only using it to get
at the platform_device pointer or struct device pointer, so fixing all
of the users was trivial.
This also makes it more difficult for device init code to directly
access members of struct omap_device, and allows for easier changing
of omap_device internals.
Cc: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Now that all of the users of the OMAP_CHIP bitfield code have been converted
to use lists, the OMAP_CHIP code, data, and declarations can be removed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
At Tony's request, remove the OMAP_CHIP* flags from the hwmod data, and
replace it instead with chip family, variant, and ES level-specific lists
of hwmods to register.
Thanks to Gražvydas Ignotas <notasas@gmail.com> for finding a bug in the
AM3517/3505 support, and for other review comments.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Gražvydas Ignotas <notasas@gmail.com>
At Tony's request, remove the omap_chip bitmasks from the powerdomain
definitions. Instead, initialize powerdomains based on one or more
lists that are applicable to a particular SoC family, variant, and
silicon revision.
Gražvydas Ignotas <notasas@gmail.com> found and reported a bug in a
related patch that also applied to this patch - thanks Gražvydas.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Gražvydas Ignotas <notasas@gmail.com>
In preparation for OMAP_CHIP() removal, split pwrdm_init() into three
functions. This allows some of them to be called multiple times: for
example, pwrdm_register_pwrdms() can be called once to register
powerdomains that are common to a group of SoCs, and once to register
powerdomains that are specific to a single SoC.
The appropriate order to call these functions - which is enforced
by the code - is:
1. pwrdm_register_platform_funcs()
2. pwrdm_register_pwrdms() (can be called multiple times)
3. pwrdm_complete_init()
Convert the OMAP2, 3, and 4 powerdomain init code to use these new
functions.
While here, improve documentation, and increase CodingStyle
conformance by shortening some local variable names.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
At Tony's request, remove the omap_chip bitmasks from the clockdomain
and clockdomain dependency definitions. Instead, initialize
clockdomains based on one or more lists that are applicable to a
particular SoC family, variant, and silicon revision.
Tony Lindgren <tony@atomide.com> found a bug in a previous version of this
patch - thanks Tony.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Tony Lindgren <tony@atomide.com>
In preparation for OMAP_CHIP() removal, split clkdm_init() into four
functions. This allows some of them to be called multiple times: for
example, clkdm_register_clkdms() can be called once to register
clockdomains that are common to a group of SoCs, and once to register
clockdomains that are specific to a single SoC.
The appropriate order to call these functions - which is enforced
by the code - is:
1. clkdm_register_platform_funcs()
2. clkdm_register_clkdms() (can be called multiple times)
3. clkdm_register_autodeps() (optional; deprecated)
4. clkdm_complete_init()
Convert the OMAP2, 3, and 4 clockdomain init code to use these new
functions.
While here, improve documentation, and increase CodingStyle
conformance by shortening some local variable names.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
omap3_cpuinfo() contains essentially duplicated code from
omap3_check_revision(), just for the purpose of determining the chip ES level.
Set the cpu_rev char array pointer in omap3_check_revision() instead,
and drop the now-useless code from omap3_cpuinfo().
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Igor Grinberg <grinberg@compulab.co.il>
Tested-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
Emit a warning to the console in omap3_check_revision() if that code
cannot determine what type of SoC the system is currently running on.
Remove some extra whitespace, remove some duplicate code, and
add an appropriate comment to a fallthrough case.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Hemant Pedanekar <hemantp@ti.com>
Tested-by: Igor Grinberg <grinberg@compulab.co.il>
Tested-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
Use explicit revision codes for OMAP/AM 3505/3517 ES levels, as the rest
of the OMAP2+ SoCs do in mach-omap2/cpu.c.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Sanjeev Premi <premi@ti.com>
Tested-by: Igor Grinberg <grinberg@compulab.co.il>
Tested-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
The OMAP3505/AM3505 appears to be based on the same silicon as the
OMAP3517/AM3517, with some features disabled via eFuse bits. Follow
the same practice as OMAP3430 and identify these devices internally as
part of the OMAP3517/AM3517 family.
The OMAP3503/3515/3525/3530 chips appear to be based on the same silicon
as the OMAP3430, with some features disabled via eFuse bits. Identify
these devices internally as part of the OMAP3430 family.
Remove the old OMAP35XX_CLASS, which actually covered two very different
chip families. The OMAP3503/3515/3525/3530 chips will now be covered by
OMAP343X_CLASS, since the silicon appears to be identical. For the
OMAP3517/AM3517 family, create a new class, OMAP3517_CLASS.
Thanks to Tony Lindgren <tony@atomide.com> for some help with the second
revision of this patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Sanjeev Premi <premi@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Tested-by: Igor Grinberg <grinberg@compulab.co.il>
Tested-by: Abhilash Koyamangalath <abhilash.kv@ti.com>
Add a missing array terminator to omap2430_usbhsotg_addrs[]. Without
this terminator, the omap_hwmod resource building code runs off the
end of the array, resulting in at least this error -- if not worse
behavior:
[ 0.578002] musb-omap2430: failed to claim resource 4
[ 0.583465] omap_device: musb-omap2430: build failed (-16)
[ 0.589294] Could not build omap_device for musb-omap2430 usb_otg_hs
This should have been part of commit
78183f3fdf ("omap_hwmod: use a null
structure record to terminate omap_hwmod_addr_space arrays") but was
evidently missed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Also, clean up error messages by adding missing whitespace, fixing
capitalisations, removing double newlines, and reducing verbosity.
Signed-off-by: Johan Hovold <jhovold@gmail.com>
[khilman@ti.com: minor changelog/subject edits]
Signed-off-by: Kevin Hilman <khilman@ti.com>
Fix misc. typos in various comment and error message.
Signed-off-by: Johan Hovold <jhovold@gmail.com>
[khilman@ti.com: minor changelog/subject edits]
Signed-off-by: Kevin Hilman <khilman@ti.com>
commit 20d5d55149 (OMAP2: PM debug:
remove register dumping) left some code lying around which doesn't
do anything. Clean it up.
Signed-off-by: Michael Jones <michael.jones@matrix-vision.de>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Introduce them for each omap variant and just make them all call
omap2_init_common_infrastructure for now. Do this for each board-*.c
file except for board-generic and board-omap3beagle as they use
the same machine ID for multiple SoCs.
No functional changes.
Signed-off-by: Tony Lindgren <tony@atomide.com>
There's no need to call omap2_init_common_devices from init_early.
It no longer does anything else except reprogram the memory timings
for some boards, so it's better to do it later so we have a chance
to get console messages if something goes wrong.
Move it to happen after omap_serial_init gets called. And while
patching it anyways, rename it to omap_sdrc_init as suggested by
Benoit Cousson <b-cousson@ti.com>.
Signed-off-by: Tony Lindgren <tony@atomide.com>
As noted by Grant Likely <grant.likely@secretlab.ca>, omap2+ Makefile unnecessarily
repeats entries for common device init code instead of using Kconfig symbol.
Remove references to hsmmc.o and board-flash.o. Also omap_phy_internal.o
references can be removed once it has some Kconfig symbol to use.
Signed-off-by: Tony Lindgren <tony@atomide.com>
The oscillator that supplies GPT12_FCLK and WDT1_FCLK exists in the
WKUP powerdomain[1]. This resolves at least one boot-time warning:
omap_hwmod: gpt12_fck: missing clockdomain for gpt12_fck.
1. _OMAP34xx Multimedia High Security (HS) Device Silicon Revision 3.1.x
Security Addendum Version K (SWPU119K)_ Figure 3-29. August 2010.
Fix the following compile warning:
arch/arm/mach-omap2/clock44xx_data.c: In function 'omap4xxx_clk_init':
arch/arm/mach-omap2/clock44xx_data.c:3371:6: warning: 'cpu_clkflg' may be used uninitialized in this function
The approach taken here is intended to work if omap4xxx_clk_init() is
converted into an initcall.
Thanks to Bjarne Steinsbo <bsteinsbo@gmail.com> for proposing another
approach.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Bjarne Steinsbo <bsteinsbo@gmail.com>
After commit 665d001338 ("OMAP2+: hwmod:
Follow the recommended PRCM module enable sequence"), device drivers
for OMAP IP blocks that do not use runtime PM can cause oopses or
kernel instability[1][2].
This is because those non-runtime PM drivers do not use the hwmod
code, which implements the correct IP block enable and disable
sequence.
Several options for dealing with this problem have been proposed:
1. Add a new field to the OMAP struct clk to mark clocks that are
currently used by non-runtime PM drivers. Modify the clock code to
use the old clockdomain sequence for these marked clocks. As
drivers are converted to use runtime PM, remove the annotation from
the clocks.
2. Similar to #1, but associate the flag with the struct omap_clk
instead.
3. Add IDLEST wait support to the OMAP4 clock code, similar to the way
it is implemented for OMAP2/3, and enable it in each struct clk
currently used by non-runtime PM drivers. As drivers are converted
to use runtime PM, remove the annotation from the clocks.
4. Do nothing; leave the problem to those responsible for the
unconverted drivers.
5. Re-enable clock-based clockdomain control in the OMAP4 clock code.
This would revert back to the behavior of Linux 3.0, simply with a
slightly longer module enable/disable latency.
Unfortunately, no approach seemed particularly good. Options 1
through 3 seemed unwise due to the following reasons:
A. The OMAP struct clks are intended primarily to describe hardware
clock nodes, and the intention is that no driver-specific data
should be stored there (applies to #1)
B. The resulting patch would have been quite large for the -rc series
(applies to #1, #2, #3)
C. The patch would have been a new, yet temporary hack; and similar fixes
have drawn negative comments in the recent past (see for example [3])
Option 4 is undesirable because commit
665d001338 ("OMAP2+: hwmod: Follow the
recommended PRCM module enable sequence") has resulted in a less
stable kernel; and kernel stability is more important than OMAP4 power
management.
Option 5 is the approach taken in this patch. This seemed to be the
least intrusive approach for 3.1-rc.
The approach in this patch was originally proposed by Ohad Ben-Cohen
<ohad@wizery.com>. I'm simply writing the commit message and passing
it along.
...
Thanks to Luciano Coelho <coelho@ti.com> for reporting the problem.
Thanks to Ohad Ben-Cohen <ohad@wizery.com> for tracking the problem
down, generating a temporary workaround, and proposing a patch to deal
with the problem. Thanks to Rajendra Nayak <rnayak@ti.com> for
proposing another patch to deal with the problem. Thanks to Felipe
Balbi <balbi@ti.com> for comments.
1. Coelho, Luciano <coelho@ti.com>. _Re: Oops on ehci_hcd when
booting 3.0.0-rc2 on panda_. Tue, 09 Aug 2011 14:26:08 +0300.
Posted to the <linux-omap@vger.kernel.org> mailing list. Available
from (among others)
http://www.spinics.net/linux/lists/linux-omap/msg55213.html
2. Munegowda, Keshava <keshava_mgowda@ti.com>. _Re: Oops on ehci_hcd
when booting 3.0.0-rc2 on panda_. Thu, 11 Aug 2011 13:51:05 +0530.
Posted to the <linux-omap@vger.kernel.org> mailing list. Available
from (among others)
http://www.spinics.net/linux/lists/linux-omap/msg55371.html
3. King, Russell <linux@arm.linux.org.uk>. _Re: [PATCH 5/8] OMAP4:
PM: TEMP: Prevent l3init from idling/force sleep_. Thu, 23 Jun
2011 16:22:49 +0100. Posted to the <linux-omap@vger.kernel.org>
mailing list. Available from (among others)
http://www.mail-archive.com/linux-omap@vger.kernel.org/msg51392.html
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Luciano Coelho <coelho@ti.com>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
While using clockdomain force wakeup method, not waiting for powerdomain
to be effectively ON may end up locking the clockdomain FSM until a
next wakeup event occurs.
One such issue was seen on OMAP4430, where L4_PER was periodically
getting stuck in in-transition state when transitioning from from OSWR to ON.
This issue was reported and investigated by Patrick Titiano <p-titiano@ti.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Reported-by: Patrick Titiano <p-titiano@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
[paul@pwsan.com: updated to apply; added transition wait on clkdm_deny_idle();
remove two superfluous pwrdm_wait_transition() calls]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Program all powerdomain target state as ON; this is to prevent domains
from hitting low power states (if bootloader has target states set to
something other than ON) and potentially even losing context while PM
is not fully initialized, which can cause the system to crash. The PM
late init code can then program the desired target state for all the
power domains.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[paul@pwsan.com: dropped comment typo hunk; fixed comment indent and moved
to kerneldoc; moved code to pwrdm_init(); changed pwrdm_init() argument name
to prevent clash; cleaned up patch description]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
If CONFIG_OMAP_32K_TIMER is not selected and dmtimer is used as clocksource, the
timer stops counting once overflow occurs as it was not set in autoreload mode.
This results into timekeeping failure: for example, 'sleep 1' at the shell after
the timer counter overflow would hang.
This patch sets up autoreload when starting the clocksource timer which fixes
the above issue.
Signed-off-by: Hemant Pedanekar <hemantp@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Since 7203f8a48b (arm: mach-omap2: remove
NULL board_mux from board files) NULL board_mux is defined in mux.h.
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Use kstrdup rather than duplicating its implementation
The semantic patch that makes this output is available
in scripts/coccinelle/api/kstrdup.cocci.
More information about semantic patching is available at
http://coccinelle.lip6.fr/
Signed-off-by: Thomas Meyer <thomas@m3y3r.de>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Builds for multi-OMAP2 (e.g., OMAP2420 with OMAP2430) with
CONFIG_ARCH_OMAP4=n fail with the following errors:
arch/arm/mach-omap2/built-in.o: In function `_enable_module':
arch/arm/mach-omap2/omap_hwmod.c:701: undefined reference to `omap4_cminst_module_enable'
arch/arm/mach-omap2/built-in.o: In function `_disable_module':
arch/arm/mach-omap2/omap_hwmod.c:726: undefined reference to `omap4_cminst_module_disable'
arch/arm/mach-omap2/built-in.o: In function `_wait_target_disable':
arch/arm/mach-omap2/omap_hwmod.c:1179: undefined reference to `omap4_cminst_wait_module_idle'
This is probably due to the preprocessor directives in
arch/arm/plat-omap/include/plat/cpu.h that convert some cpu_is_omap*()
expressions from preprocessor directives into something that is only
resolvable during runtime, if multiple OMAP2 build targets are
selected.
Thanks to Tony Lindgren <tony@atomide.com> for reporting.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Commit b22f954 (OMAP4: Move common twl6030 configuration to twl-common)
caused compile failures for code for OMAP arch which is not selected by
the config.
Fixes issues like:
With CONFIG_ARCH_OMAP3=y and CONFIG_ARCH_OMAP4=n, I'm getting this:
arch/arm/mach-omap2/built-in.o:(.data+0xf99c): undefined reference to `omap4430_phy_init'
arch/arm/mach-omap2/built-in.o:(.data+0xf9a0): undefined reference to `omap4430_phy_exit'
arch/arm/mach-omap2/built-in.o:(.data+0xf9a4): undefined reference to `omap4430_phy_power'
arch/arm/mach-omap2/built-in.o:(.data+0xf9a8): undefined reference to `omap4430_phy_set_clk'
arch/arm/mach-omap2/built-in.o:(.data+0xf9ac): undefined reference to `omap4430_phy_suspend'
Fix the problem by moving the code to ifdef sections for omap3 and omap4.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
[tony@atomide.com: updated comments]
Signed-off-by: Tony Lindgren <tony@atomide.com>
omap_sr_disable_reset_volt is called with irqs off in omapx_enter_sleep,
as part of idle sequence, this eventually calls sr_disable and
pm_runtime_put_sync. pm_runtime_put_sync calls rpm_idle, which will
enable interrupts in order to call the callback. In this short interval
when interrupts are enabled, scenarios such as the following can occur:
while interrupts are enabled, the timer interrupt that is supposed to
wake the device out of idle occurs and is acked, so when the CPU finally
goes to off, the timer is already gone, missing a wakeup event.
Further, as the documentation for runtime states:"
However, subsystems can use the pm_runtime_irq_safe() helper function
to tell the PM core that a device's ->runtime_suspend() and ->runtime_resume()
callbacks should be invoked in atomic context with interrupts disabled
(->runtime_idle() is still invoked the default way)."
Hence, replace pm_runtime_put_sync with pm_runtime_put_sync_suspend
to invoke the suspend handler and shut off the fclk for SmartReflex
module instead of using the idle handler in interrupt disabled context.
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Colin Cross <ccross@google.com>
[khilman@ti.com: minor Subject edits]
Signed-off-by: Kevin Hilman <khilman@ti.com>
Board code should not touch omap_device internals. To get the MPU/IVA devices,
use existing APIs: omap2_get_mpu_device(), omap2_get_iva_device().
Signed-off-by: Kevin Hilman <khilman@ti.com>
SmartReflex should be disabled while entering low power mode due to
a) SmartReflex values are not defined for retention voltage, further
b) with SmartReflex enabled, if CPU enters lower c-states, FSM will try
to bump the voltage to current OPP's voltage for which it has entered c-state;
hence SmartReflex needs to be disabled for MPU, CORE and IVA voltage
domains in idle path before enabling auto retention voltage achievement
on the device.
However, since the current pm_runtime setup for SmartReflex devices are
setup to allow callbacks to be invoked with interrupts enabled, calling
SmartReflex enable/disable from other contexts such as idle paths
where preemption is disabled causes warnings such as the following
indicating of a potential race.
[ 82.023895] [<c04d079c>] (__irq_svc+0x3c/0x120) from [<c04d0484>] (_raw_spin_unlock_irq+0x28/0x2c)
[ 82.023895] [<c04d0484>] (_raw_spin_unlock_irq+0x28/0x2c) from [<c0323234>] (rpm_callback+0x4c/0x68)
[ 82.023956] [<c0323234>] (rpm_callback+0x4c/0x68) from [<c0323f7c>] (rpm_resume+0x338/0x53c)
[ 82.023956] [<c0323f7c>] (rpm_resume+0x338/0x53c) from [<c03243f4>] (__pm_runtime_resume+0x48/0x60)
[ 82.023986] [<c03243f4>] (__pm_runtime_resume+0x48/0x60) from [<c008aee0>] (sr_enable+0xa8/0x19c)
[ 82.023986] [<c008aee0>] (sr_enable+0xa8/0x19c) from [<c008b2fc>] (omap_sr_enable+0x50/0x90)
[ 82.024017] [<c008b2fc>] (omap_sr_enable+0x50/0x90) from [<c00888c0>] (omap4_enter_sleep+0x138/0x168)
Instead, we use pm_runtime_irq_safe to tell the PM core that callbacks can be
invoked in interrupt disabled contexts.
Acked-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
[khilman@ti.com: minor changelog edits]
Signed-off-by: Kevin Hilman <khilman@ti.com>
CONFIG_PM is no longer a user-selectable Kconfig option. Rather it is
automatically enabled if either CONFIG_SUSPEND or CONFIG_RUNTIME_PM is
enabled, so having a 'select PM' here is redunant when 'select
CONFIG_PM_RUNTIME' is present.
Signed-off-by: Kevin Hilman <khilman@ti.com>
Benoit Cousson