Modify the Cortex-M SysTick driver to work when the SoC resets SysTick
in low-power mode. This adds an invisible Kconfig symbol that must be
selected by SoCs with such behavior.
Signed-off-by: Mathieu Choplain <mathieu.choplain@st.com>
Add support for a hook-based low-power mode timer to the Cortex-M SysTick
driver, in addition to the existing Counter API-based timer. This is useful
on platforms such as the STM32WB0 where the wake-up timer is too limited
for a Counter API driver to be implemented on top of it.
Signed-off-by: Mathieu Choplain <mathieu.choplain@st.com>
The Cortex-M SysTick supported collaboration with a so-called "IDLE timer"
to work even in low-power modes, when the SoC may disable its clock.
Rename "IDLE timer" to "low-power mode timer" to better reflect what this
timer is used for, and transform the single option into a Kconfig choice
to allow other behaviors to be implemented and chosen.
Signed-off-by: Mathieu Choplain <mathieu.choplain@st.com>
The current driver implements the global defined "systick" interrupt
callback "sys_clock_isr" as a standard C function with an argument.
However, ARM's direct interrupt handlers do not have any arguments;
they must be declared as "void handler(void)".
Additionally, the direct handler should include some missing special
header/footer.
Fixes: #75693
Signed-off-by: Andrey VOLKOV <andrey.volkov@munic.io>
Replace useless "TICKLESS" define by
IS_ENABLED(CONFIG_TICKLESS_KERNEL) one, and replace error-prone
'cycle_t' pseudotype defs by real typedefs.
Signed-off-by: Andrey VOLKOV <andrey.volkov@munic.io>
SysTick usually has higher measurement resolution than the IDLE timer.
When the time in low power mode is very short or 0, it is possible that
SysTick usually has measures more time since the sys_clock_set_timeout
than the idle timer.
Handle that case to keep uptime correct.
Signed-off-by: Dawid Niedzwiecki <dawidn@google.com>
The idle timer has its max value and can overflow. We measure time passed
since the sys_clock_set_timeout call. Take possibility of the overflow
into account.
Signed-off-by: Dawid Niedzwiecki <dawidn@google.com>
When the idle timer is in use, we calculate number of cycles passed
since the sys_clock_set_timeout call.
The cycle counter can overflow easily, when the counter is 32-bit wide.
Handle that case.
Signed-off-by: Dawid Niedzwiecki <dawidn@google.com>
Some chips, that use Cortex-M SysTick as the system timer, disable a
clock in a low power mode, that is the input for the SysTick e.g.
STM32Fx family.
It blocks enabling power management for these chips. The wake-up
function doesn't work and the time measurement is lost.
Add an additional IDLE timer that handles these functionality when the
system is about to enter IDLE. It has to wake up the chip and update the
cycle counter by time not measured by the SysTick. The IDLE timer has to
support counter API (setting alarm and reading current value).
Signed-off-by: Dawid Niedzwiecki <dawidn@google.com>
`last_load` is the full N cycles and `SysTick->LOAD` should
be loaded with `last_load - 1` for the calculations work
correctly.
Note: This only affects a kernel in ticked operation.
Tickless kernels periodically restart the timer correctly.
Signed-off-by: Grant Ramsay <gramsay@enphaseenergy.com>
The existing implementation did not properly
handle when `SysTick->VAL` is zero.
This caused three subtle edge cases:
* val1=0,COUNTFLAG=0,val2=0
This should result in no cycles elapsed,
however `(last_load - val2) = last_load`.
So an extra `last_load` cycles was returned.
* val1=0,COUNTFLAG=0,val2=(last_load-1)
This should result in 1 cycle elapsed,
however `val1 < val2` so an extra `last_load`
cycles was returned.
* val1=[2,1,0],COUNTFLAG=1,val2=0
This should result in `last_load` cycles elapsed.
However, `last_load * 2` cycles was returned.
To fix the calculation, val1 and val2 are first
wrapped/realigned from [0:last_load-1] to [1:last_load].
Tidy comments to better reflect the SysTick
behaviour and link reference manuals.
Signed-off-by: Grant Ramsay <gramsay@enphaseenergy.com>
Timer "drivers" do not use the device model infrastructure, they are
singletons with a SYS_INIT call. This means they do not have to include
device.h but init.h. Things worked because device.h includes init.h.
Signed-off-by: Gerard Marull-Paretas <gerard@teslabs.com>
The CMSIS module glue code was part of arch/ directory. Move it to
modules/cmsis, and provide a single entry point for it: cmsis_core.h.
This entry header will include the right CMSIS header (M or A/R).
To make this change possible, CMSIS module Kconfig/CMake are declared as
external, allowing us to add a new Zephyr include directory.
All files including CMSIS have been updated.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
This driver, due to its limited 24-bits counter, is already tracking a
cycle count in software. Allow that count to be a 64-bits value so this
won't wrap in a matter of only a few seconds when the hardware clock
is fast.
This is very cheap to do as expensive math operations (i.e. divisions)
are performed only on counter intervals whose values fit in 32 bits like
before.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Many areas of Zephyr divide and round up without using the DIV_ROUND_UP
macro. Make use of it, so that we make use of a tested system macro and
at the same time we make code more readable.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The init infrastructure, found in `init.h`, is currently used by:
- `SYS_INIT`: to call functions before `main`
- `DEVICE_*`: to initialize devices
They are all sorted according to an initialization level + a priority.
`SYS_INIT` calls are really orthogonal to devices, however, the required
function signature requires a `const struct device *dev` as a first
argument. The only reason for that is because the same init machinery is
used by devices, so we have something like:
```c
struct init_entry {
int (*init)(const struct device *dev);
/* only set by DEVICE_*, otherwise NULL */
const struct device *dev;
}
```
As a result, we end up with such weird/ugly pattern:
```c
static int my_init(const struct device *dev)
{
/* always NULL! add ARG_UNUSED to avoid compiler warning */
ARG_UNUSED(dev);
...
}
```
This is really a result of poor internals isolation. This patch proposes
a to make init entries more flexible so that they can accept sytem
initialization calls like this:
```c
static int my_init(void)
{
...
}
```
This is achieved using a union:
```c
union init_function {
/* for SYS_INIT, used when init_entry.dev == NULL */
int (*sys)(void);
/* for DEVICE*, used when init_entry.dev != NULL */
int (*dev)(const struct device *dev);
};
struct init_entry {
/* stores init function (either for SYS_INIT or DEVICE*)
union init_function init_fn;
/* stores device pointer for DEVICE*, NULL for SYS_INIT. Allows
* to know which union entry to call.
*/
const struct device *dev;
}
```
This solution **does not increase ROM usage**, and allows to offer clean
public APIs for both SYS_INIT and DEVICE*. Note that however, init
machinery keeps a coupling with devices.
**NOTE**: This is a breaking change! All `SYS_INIT` functions will need
to be converted to the new signature. See the script offered in the
following commit.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
init: convert SYS_INIT functions to the new signature
Conversion scripted using scripts/utils/migrate_sys_init.py.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
manifest: update projects for SYS_INIT changes
Update modules with updated SYS_INIT calls:
- hal_ti
- lvgl
- sof
- TraceRecorderSource
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
tests: devicetree: devices: adjust test
Adjust test according to the recently introduced SYS_INIT
infrastructure.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
tests: kernel: threads: adjust SYS_INIT call
Adjust to the new signature: int (*init_fn)(void);
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
`#defines` do NOT sepecify a type. They will either adopt a native
system type or type of the value that was passed into the expression.
This can lead to warnings such as, "warning: comparison of integer
expressions of different signedness: 'uint32_t' {aka 'unsigned int'}
and 'int' [-Wsign-compare]".
By casting expressions, such as `MAX_TICKS` to `k_ticks_t`, we can
force the appropriate types and resolve these warnings.
Signed-off-by: Zachary J. Fields <zachary_fields@yahoo.com>
Change automated searching for files using "IRQ_CONNECT()" API not
including <zephyr/irq.h>.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
In order to bring consistency in-tree, migrate all drivers to the new
prefix <zephyr/...>. Note that the conversion has been scripted, refer
to #45388 for more details.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The weak symbol sys_clock_driver_init has been removed, therefore moving
the init responsability to the drivers themselves. As a result, the init
function has now been made static on all drivers and moved to the
bottom, following the convention used in other areas.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
With this patch the sys_clock_set_timeout function counts the cycles
elapsed while computing the systick timer's new load (tickless mode).
This cycles are then added to the total cycle count instead of being
lost.
This patch mitigates uptime drifting in tickless mode (especially when
high frequency timers are registered).
Signed-off-by: Simon Guinot <simon.guinot@seagate.com>
This feature predated the tickless kernel and has been in legacy mode
for a while. We now have no drivers or systems that do not support
tickless, so remove this option and cleanup the code to only use
tickless.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
This is another API that is being used in all timer drivers and is not
internal to the clock subsystem. Remove the leading z_ and make promote
it to a cross-subsystem API.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The clock/timer APIs are not application facing APIs, however, similar
to arch_ and a few other APIs they are available to implement drivers
and add support for new hardware and are documented and available to be
used outside of the clock/kernel subsystems.
Remove the leading z_ and provide them as clock_* APIs for someone
writing a new timer driver to use.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Replaces all existing variants of value clamping with the MIN and MAX
macros with the CLAMP macro.
Signed-off-by: Trond Einar Snekvik <Trond.Einar.Snekvik@nordicsemi.no>
Now that device_api attribute is unmodified at runtime, as well as all
the other attributes, it is possible to switch all device driver
instance to be constant.
A coccinelle rule is used for this:
@r_const_dev_1
disable optional_qualifier
@
@@
-struct device *
+const struct device *
@r_const_dev_2
disable optional_qualifier
@
@@
-struct device * const
+const struct device *
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Marking as unused (ARG_UNUSED) the parameter device in the
initialization function z_clock_driver_init when it is not used.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
In the ARM Cortex-M architecture implementation, the concepts of
"exceptions" and "interrupts" are interchangeable; whereas, in the
Cortex-A/-R architecture implementation, they are considered separate
and therefore handled differently (i.e. `z_arm_exc_exit` cannot be used
to exit an "interrupt").
This commit fixes all `z_arm_exc_exit` usages in the interrupt handlers
to use `z_arm_int_exit`.
NOTE: In terms of the ARM AArch32 Cortex-A and Cortex-R architecture
implementations, the "exceptions" refer to the "Undefined
Instruction (UNDEF)" and "Prefetch/Data Abort (PABT/DABT)"
exceptions, while "interrupts" refer to the "Interrupt (IRQ)",
"Fast Interrupt (FIQ)" and "Software Interrupt/Supervisor Call
(SWI/SVC)".
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Add a k_timeout_t type, and use it everywhere that kernel API
functions were accepting a millisecond timeout argument. Instead of
forcing milliseconds everywhere (which are often not integrally
representable as system ticks), do the conversion to ticks at the
point where the timeout is created. This avoids an extra unit
conversion in some application code, and allows us to express the
timeout in units other than milliseconds to achieve greater precision.
The existing K_MSEC() et. al. macros now return initializers for a
k_timeout_t.
The K_NO_WAIT and K_FOREVER constants have now become k_timeout_t
values, which means they cannot be operated on as integers.
Applications which have their own APIs that need to inspect these
vs. user-provided timeouts can now use a K_TIMEOUT_EQ() predicate to
test for equality.
Timer drivers, which receive an integer tick count in ther
z_clock_set_timeout() functions, now use the integer-valued
K_TICKS_FOREVER constant instead of K_FOREVER.
For the initial release, to preserve source compatibility, a
CONFIG_LEGACY_TIMEOUT_API kconfig is provided. When true, the
k_timeout_t will remain a compatible 32 bit value that will work with
any legacy Zephyr application.
Some subsystems present timeout (or timeout-like) values to their own
users as APIs that would re-use the kernel's own constants and
conventions. These will require some minor design work to adapt to
the new scheme (in most cases just using k_timeout_t directly in their
own API), and they have not been changed in this patch, instead
selecting CONFIG_LEGACY_TIMEOUT_API via kconfig. These subsystems
include: CAN Bus, the Microbit display driver, I2S, LoRa modem
drivers, the UART Async API, Video hardware drivers, the console
subsystem, and the network buffer abstraction.
k_sleep() now takes a k_timeout_t argument, with a k_msleep() variant
provided that works identically to the original API.
Most of the changes here are just type/configuration management and
documentation, but there are logic changes in mempool, where a loop
that used a timeout numerically has been reworked using a new
z_timeout_end_calc() predicate. Also in queue.c, a (when POLL was
enabled) a similar loop was needlessly used to try to retry the
k_poll() call after a spurious failure. But k_poll() does not fail
spuriously, so the loop was removed.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Before introducing the code for ARM64 (AArch64) we need to relocate the
current ARM code to a new AArch32 sub-directory. For now we can assume
that no code is shared between ARM and ARM64.
There are no functional changes. The code is moved to the new location
and the file paths are fixed to reflect this change.
Signed-off-by: Carlo Caione <ccaione@baylibre.com>
Similar to what we do in other timer drivers, the maximum ticks
supplied in z_clock_set_timeout(..) needs to be MAX_TICKS at
maximum, when K_FOREVER is supplied as argument to the function.
In addition to that, the value we load onto the SysTick LOAD
register shall be truncated to MAX_CYCLES. This is required
to prevent loading a trash value to LOAD register, as only
the lowest 24 bits may be safely written.
Finally, we move the enforcement of the minimum delay to be
programmed on LOAD (i.e. MIN_DELAY) at the end step of the
calculation of the cycles-to-be-programmed. This prevents
from misscalculating the delay, as any required adjustment
is applied at the end, after the delay is rounded up to
the next tick boundary.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
When setting a timeout measure the number of accumulated unannounced
ticks. If this value exceeds half the 32-bit cycle counter range
force an announcement so the unannounced cycles are incorporated into
the system tick counter.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The commit fixes the update of the absolute counter of HW cycles
in the SysTick ISR for TICKLESS mode.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
The previous solution depended on a magic number and was inefficient
(entered the second-wrap conditional even when a second wrap hadn't
been observed). Replace with an algorithm that is deterministic.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Add detailed documentation for the internal 'elapsed()'
function, as well as for the local counter variables used
in the SysTick driver.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Unsupported bits of the Current Value Register
are read as zero, so we remove the redundant
ANDing with the max supported counter value.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Removed workarounds in systick driver as they prevent normal usage in
TICKLESS systems. Driver still behaved like an interrupt based ticker.
Signed-off-by: Andrei Gansari <andrei.gansari@nxp.com>
This patch re-namespaces global variables and functions
that are used only within the arch/arm/ code to be
prefixed with z_arm_.
Some instances of CamelCase have been corrected.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
On some SoCs the frequency of the system clock is obtained at run time
as the exact configuration of the hardware is not known at compile time.
On such platforms using CONFIG_SYS_CLOCK_HW_CYCLES_PER_SEC define
directly introduces timing errors.
This commit replaces CONFIG_SYS_CLOCK_HW_CYCLES_PER_SEC by the call
to inline function sys_clock_hw_cycles_per_sec() which always returns
correct frequency of the system clock.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
Move internal and architecture specific headers from include/drivers to
subfolder for timer:
include/drivers/timer
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Clearing the pending IRQs when resetting the timeout fixes the
forward time drifting, but the change needs more investigation
until we are sure this won't break kernel time management.
Reverting the change to get 1.14 release out.
This reverts commit 2895da02a4.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
In the unlucky scenario of a SysTick event (wrap) occurring
while we re-program the last_load value, the SysTick ISR
will run immediately after we unlock interrupts. In that
case the timeout we have just configured will expire
instantaneously, leading to operations being executed
much earlier than expected. Avoid this by clearing possibly
pending SysTick exceptions (writing 1 to ICSR.PENDSTCLR).
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
When the counter reaches zero, it reloads the value in
SYST_RVR on the next clock edge. This means that if the
LOAD value is N, the interrupt ("tick") is triggered
every N+1 cycles. Therefore, when we operate in tickless
mode, and we want to schedule the next timeout, we need
to configure the LOAD value with last_load - 1, in order
to get an event in last_load cycles.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
When the counter reaches zero, it reloads the value in
SYST_RVR on the next clock edge. This means that if the
LOAD value is N, the interrupt ("tick") is triggered
every N+1 cycles. Therefore, when we operate in non-
tickless mode, we need to configure the LOAD value
with CYC_PER_TICK - 1, in order to get an event
every CYC_PER_TICK cycles.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
The SysTick logic looked logically sound, but it was allowing us to
set a LOAD value as low as 512 cycles. On other platforms, that
minimum future interrupt delay is there to protect the "read, compute,
write, unmask" cycle that sets the new interrupt from trying to set
one sooner than it can handle.
But with SysTick, that value then becomes the value of the LOAD
register, which is effectively the frequency with which timer
interrupts arrive. This has two side effects:
1. It opens up the possibility that future code that masks interrupts
longer than 512 cycles will miss more than one overflow, slipping
the clock backward as viewed by z_clock_announce().
2. The original code only understood one overflow cycle, so in the
event we do set one of these very near timeouts and then mask
interrupts, we'll only add at most one overflow to the "elapsed()"
time, slipping the CURRENT time backward (actually turning it into
a non-monotonic sawtooth which would slip every LOAD cycle) and
thus messing up future scheduled interrupts, slipping those forward
relative to what the ISR was counting.
This patch simplifies the logic for reading SysTick VAL/CTRL (the loop
wasn't needed), handles the case where we see more than one overflow,
and increases the MIN_DELAY cycles from 512 to 1/16th of a tick.
Fixes#15216
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Mathieu Choplain
WenBin Zhang
Andrey VOLKOV
Dawid Niedzwiecki
Grant Ramsay
Gerard Marull-Paretas
Gerard Marull-Paretas
Nicolas Pitre
Zachary J. Fields
Simon Guinot
Anas Nashif
Trond Einar Snekvik
Tomasz Bursztyka
Flavio Ceolin
Kumar Gala
Stephanos Ioannidis
Andy Ross
Carlo Caione
Ioannis Glaropoulos
Peter Bigot
Andrei Gansari
Andrew Boie
Piotr Zięcik