As suggested, this table is an optimized version for rx51 and we
remove it in order to consolidate the cpuidle code between omap3
and omap4, we remove this specific data definition which is used
to override the default omap3 latencies but at the cost of extra
code and complexity.
In order to not lose the values which probably took time to be
measured, the table is converted into a comment with an array
description.
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Use core cpuidle timekeeping and irqen wrapper and remove that
handling from this code.
Signed-off-by: Robert Lee <rob.lee@linaro.org>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Reviewed-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Tested-by: Jean Pihet <j-pihet@ti.com>
Acked-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Omap_uart_can_sleep function blocks system wide low power state until
uart is active remove this func and add qos requests to prevent
MPU from transitioning.
Keep qos request to default value which will allow MPU to transition
and while uart baud rate is available calculate the latency value
from the baudrate and use the same to hold constraint while uart clocks
are enabled, and if uart is auto-idled the constraint is updated with
default constraint value allowing MPU to transition.
Qos requests are blocking notifier calls so put these requests to
work queue, also the driver uses irq_safe version of runtime API's
and callbacks can be called in interrupt disabled context.
So to avoid warn on slow path warning while using qos update
API's from runtime callbacks use the qos_work_queue.
During bootup the runtime_resume call backs might not be called and runtime
callback gets called only after uart is idled by setting the autosuspend
timeout. So qos_request from runtime resume callback might not activated during
boot if uart baudrate is calculated during bootup for console uart, so schedule
the qos_work queue once we calc_latency while configuring the uart port.
Flush and complete any pending qos jobs in work queue while suspending.
Signed-off-by: Govindraj.R <govindraj.raja@ti.com>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de> (for drivers/tty changes)
Signed-off-by: Kevin Hilman <khilman@ti.com>
Save VFP CPU context using CPU PM notifier chain. VFP context
is lost when CPU hits OFF state.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Tested-by: Vishwanath BS <vishwanath.bs@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Fix the below build break by including common.h
arch/arm/mach-omap2/cpuidle34xx.c: In function 'omap3_enter_idle':
arch/arm/mach-omap2/cpuidle34xx.c:117: error: implicit declaration of function 'omap_irq_pending'
make[1]: *** [arch/arm/mach-omap2/cpuidle34xx.o] Error 1
make: *** [arch/arm/mach-omap2] Error 2
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This patch makes the cpuidle_states structure global (single copy)
instead of per-cpu. The statistics needed on per-cpu basis
by the governor are kept per-cpu. This simplifies the cpuidle
subsystem as state registration is done by single cpu only.
Having single copy of cpuidle_states saves memory. Rare case
of asymmetric C-states can be handled within the cpuidle driver
and architectures such as POWER do not have asymmetric C-states.
Having single/global registration of all the idle states,
dynamic C-state transitions on x86 are handled by
the boot cpu. Here, the boot cpu would disable all the devices,
re-populate the states and later enable all the devices,
irrespective of the cpu that would receive the notification first.
Reference:
https://lkml.org/lkml/2011/4/25/83
Signed-off-by: Deepthi Dharwar <deepthi@linux.vnet.ibm.com>
Signed-off-by: Trinabh Gupta <g.trinabh@gmail.com>
Tested-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This is the first step towards global registration of cpuidle
states. The statistics used primarily by the governor are per-cpu
and have to be split from rest of the fields inside cpuidle_state,
which would be made global i.e. single copy. The driver_data field
is also per-cpu and moved.
Signed-off-by: Deepthi Dharwar <deepthi@linux.vnet.ibm.com>
Signed-off-by: Trinabh Gupta <g.trinabh@gmail.com>
Tested-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Cpuidle governor only suggests the state to enter using the
governor->select() interface, but allows the low level driver to
override the recommended state. The actual entered state
may be different because of software or hardware demotion. Software
demotion is done by the back-end cpuidle driver and can be accounted
correctly. Current cpuidle code uses last_state field to capture the
actual state entered and based on that updates the statistics for the
state entered.
Ideally the driver enter routine should update the counters,
and it should return the state actually entered rather than the time
spent there. The generic cpuidle code should simply handle where
the counters live in the sysfs namespace, not updating the counters.
Reference:
https://lkml.org/lkml/2011/3/25/52
Signed-off-by: Deepthi Dharwar <deepthi@linux.vnet.ibm.com>
Signed-off-by: Trinabh Gupta <g.trinabh@gmail.com>
Tested-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The achievable power modes of the power domains in cpuidle
depends on the system wide 'enable_off_mode' knob in debugfs.
Upon changing enable_off_mode, do not change the C-states
'valid' field but instead dynamically restrict the power modes
when entering idle.
The C-states 'valid' field is just used to enable/disable some
C-states at init and shall not be changed later on.
Signed-off-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
- fix single and multi-lines comments format
- removed the omap3_idle_bm_check function and replaced the test
in omap3_enter_idle_bm by the equivalent code
- re-organize omap3_enter_idle_bm code path, assign local variables
only when needed
- reword some comments
Signed-off-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
The current implementation defines an internal structure and a
C-states array. Using those structures is redundant to the
structs used by the cpuidle framework.
This patch provides a clean-up of the internal struct, removes the
internal C-states array, stores the data using the existing cpuidle
per C-state struct and registers the mach specific data to cpuidle
C-state driver_data (accessed using cpuidle_[gs]et_statedata).
Also removes unused macros, fields and code and compacts the repeating
code using an inline helper function.
The result is more compact and more readable code as well as
reduced data RAM usage.
Also retain C1 as the only always valid C-state and system safe state.
Signed-off-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
- sleep_latency and wake_latency are not used, replace them by
exit_latency which is used by cpuidle. exit_latency simply is
the sum of sleep_latency and wake_latency,
- replace threshold by target_residency,
- changed the OMAP3 specific cpuidle code accordingly,
- changed the OMAP3 board code accordingly.
Signed-off-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Add a description field to each idle C-state. This helps to give
better data with PowerTop and one don't have to refer to the code
to link what Cx means from system point of view while analysing
PowerTop data.
No functional change.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
In case in user has a OMAP3630 < ES1.2 the kernel should warn the user
about the ERRATUM, but using pr_warn instead of WARN_ON is already
enough, as there is nothing else the user can do besides changing the
board.
Signed-off-by: Ricardo Salveti de Araujo <ricardo.salveti@canonical.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Define the following architecture specific funtions for omap2/3/4
.clkdm_allow_idle
.clkdm_deny_idle
Convert the platform-independent framework to call these functions.
Also rename the api's by removing the omap2_ preamble.
Hence call omap2_clkdm_allow_idle as clkdm_allow_idle and
omap2_clkdm_deny_idle as clkdm_deny_idle.
Make the _clkdm_add_autodeps and _clkdm_del_autodeps as non-static
so they can be accessed from OMAP2/3 platform specific code.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
* 'idle-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-idle-2.6:
cpuidle/x86/perf: fix power:cpu_idle double end events and throw cpu_idle events from the cpuidle layer
intel_idle: open broadcast clock event
cpuidle: CPUIDLE_FLAG_CHECK_BM is omap3_idle specific
cpuidle: CPUIDLE_FLAG_TLB_FLUSHED is specific to intel_idle
cpuidle: delete unused CPUIDLE_FLAG_SHALLOW, BALANCED, DEEP definitions
SH, cpuidle: delete use of NOP CPUIDLE_FLAGS_SHALLOW
cpuidle: delete NOP CPUIDLE_FLAG_POLL
ACPI: processor_idle: delete use of NOP CPUIDLE_FLAGs
cpuidle: Rename X86 specific idle poll state[0] from C0 to POLL
ACPI, intel_idle: Cleanup idle= internal variables
cpuidle: Make cpuidle_enable_device() call poll_idle_init()
intel_idle: update Sandy Bridge core C-state residency targets
The OMAP powerdomain code and data is all OMAP2+-specific. This seems
unlikely to change any time soon. Move plat-omap/include/plat/powerdomain.h
to mach-omap2/powerdomain.h. The primary point of doing this is to remove
the temptation for unrelated upper-layer code to access powerdomain code
and data directly.
As part of this process, remove the references to powerdomain data
from the GPIO "driver" and the OMAP PM no-op layer, both in plat-omap.
Change the DSPBridge code to point to the new location for the
powerdomain headers. The DSPBridge code should not be including the
powerdomain headers; these should be removed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
Cc: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
The OMAP clockdomain code and data is all OMAP2+-specific. This seems
unlikely to change any time soon. Move plat-omap/include/plat/clockdomain.h
to mach-omap2/clockdomain.h. The primary point of doing this is to remove
the temptation for unrelated upper-layer code to access clockdomain code
and data directly.
DSPBridge also uses the clockdomain headers for some reason, so,
modify it also. The DSPBridge code should not be including the
clockdomain headers; these should be removed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
Cc: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Tested-by: Rajendra Nayak <rnayak@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Limitation i583: Self_Refresh Exit issue after OFF mode
Issue:
When device is waking up from OFF mode, then SDRC state machine sends
inappropriate sequence violating JEDEC standards.
Impact:
OMAP3630 < ES1.2 is impacted as follows depending on the platform:
CS0: for 38.4MHz as internal sysclk, DDR content seen to be stable, while
for all other sysclk frequencies, varied levels of instability
seen based on varied parameters.
CS1: impacted
This patch takes option #3 as recommended by the Silicon erratum:
Avoid core power domain transitioning to OFF mode. Power consumption
impact is expected in this case.
To do this, we route core OFF requests to RET request on the impacted
revisions of silicon.
Acked-by: Jean Pihet <j-pihet@ti.com>
[nm@ti.com: rebased the code to 2.6.37-rc2- short circuit code changed a bit]
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Eduardo Valentin <eduardo.valentin@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Currently omap3_cpuidle_update_states makes whole sale decision
on which C states to update based on enable_off_mode variable
Instead, achieve the same functionality by independently providing
mpu and core deepest states the system is allowed to achieve and
update the idle states accordingly.
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Jean Pihet <j-pihet@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
[khilman: fixed additional user of this API in OMAP CPUidle driver]
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Only OMAP2+ platforms have the System Control Module (SCM) IP block.
In the past, we've kept the SCM header file in plat-omap. This has
led to abuse - device drivers including it; includes being added that
create implicit dependencies on OMAP2+ builds; etc.
In response, move the SCM headers into mach-omap2/.
As part of this, remove the direct SCM access from the OMAP UDC
driver. It was clearly broken. The UDC code needs an indepth review for
use on OMAP2+ chips.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Cory Maccarrone <darkstar6262@gmail.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
When checking how to program the next powerstate for the PER
powerdomain, the next state of PER powerdomain was written twice.
Remove the duplicate write.
Reported-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
In an effort to simplify the core idle path, move any device-specific
special case handling from the core PM idle path into the CPUidle
pre-idle checking path.
This keeps the core, interrupts-disabled idle path streamlined and
independent of any device-specific handling, and also allows CPUidle
to do the checking only for certain C-states as needed. This patch
has the device checks in place for all states with the CHECK_BM flag,
namely all states >= C2.
This patch was inspired by a similar patch written by Tero Kristo as
part of a larger series to add INACTIVE state support.
NOTE: This is a baby-step towards decoupling device idle (or system
idle) from CPU idle. Eventually, CPUidle should only manage the CPU,
and device/system idle should be managed elsewhere.
Cc: Tero Kristo <tero.kristo@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Different boards benefit differently from the available
seven C-states for cpu idle. In most cases, only few,
properly spaced (in terms of consumption and latency)
C-states are required to make the power management
optimal. Hence we need a possibility to pass which
C-states are actually used for each board.
So added the valid field to cpuidle_params and added
support to 3430sdp, which uses the paramenter passing.
Signed-off-by: Kalle Jokiniemi <kalle.jokiniemi@digia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
The CPUidle C state latencies and thresholds are dependent on various
board specific details. This patch makes it possible to configure
these values from the respective board files.
omap3_pm_init_cpuidle() can now be optionally called from board files
to pass board specific cpuidle parameters. If the board files do not
use this function to pass the params default values are used which
might cause higher consumption dur to wrong state selection by the
governor.
This patch only updates the 3430sdp board files to use
omap3_pm_init_cpuidle().
From Kalle, in addition to original patch from Rajendra:
Building without CONFIG_CPU_IDLE or CONFIG_PM causes build to fail if
cpu idle parameters are tried to pass using omap3_pm_init_cpuidle
function.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Kalle Jokiniemi <kalle.jokiniemi@digia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
When 'enable_off_mode' is 0, the target power state for MPU
and CORE was locally changed to PWRDM_POWER_RET but, the
statistics are updated for idle state originally selected
by the governor.
This patch 'invalidates' the idle states that lead either of
MPU or Core to PWRDM_POWER_OFF state when 'enable_off_mode'
is '0'. The states are valid once 'enable_off_mode' is set
to '1'.
Added function next_valid_state() to check if current state
is valid; else get the next valid state. It is called from
omap3_enter_idle_bm().
Signed-off-by: Sanjeev Premi <premi@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Previously used u32 as temporary data storage that wraps around at 4.294s.
Signed-off-by: Tero Kristo <tero.kristo@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Fixes a bug where scheduling is delayed until next wakeup due to race
condition (e.g. interrupt requests scheduling just before omap_sram_idle
is entered.)
Signed-off-by: Tero Kristo <tero.kristo@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
It is more efficient to use pwrdm_set_next_pwrst for mpu, core and neon
instead of set_pwrdm_state in idle loop. It is anyway known that those are
active in idle loop. So no need to use set_pwrdm_state.
Signed-off-by: Jouni Hogander <jouni.hogander@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
The current definition of C-states starts from base 1.
Whereas, the cpuidle driver uses base 0. This patch
eliminates need for explicit mapping (add/ sbutract)
due to different base values.
Signed-off-by: Sanjeev Premi <premi@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
This patch introduces a new C state which allows MPU to go to WFI but keeps
the core domain active. This offers a much better wakeup latency (3us vs
10s of us for the current C1) at the cost of a higher power consumption.
Signed-off-by: Peter 'p2' De Schrijver <peter.de-schrijver@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Previously omap3_idle_init() was called in device_init, while
omap_pm_init() is called at late_initcall. This causes the cpu idle
driver to call omap_sram_idle before it is properly initialized. This
patch fixes the issue by moving omap3_idle_init into omap3_pm_init.
Signed-off-by: Kalle Jokiniemi <ext-kalle.jokiniemi@nokia.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Use the activity check for states C2 and C3 as well. This is
primarily to prevent deeper states during UART activity.
Also, if a different state is chosen than the target state, update the
'last_state' accordingly so that CPUidle state accounting is coorect.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
This patch adds support and enables state C4(MPU RET + CORE RET)
and MPU OFF states (C3 and C5.)
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Basic CPUidle driver for OMAP3 with deepest sleep state supported
being MPU CSWR.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>