Instead of obscure linker-file hacks providing dummy definitions before
we have the proper generated k-object lookup functions, provide weak
definitions that will be taken out by the real generated implementations
when they are available.
Signed-off-by: Björn Bergman <bjorn.bergman@iar.com>
Most kernel objects should be initialized well before being
manipulated in ISR context.
Event objects are no exception. Initializing a k_event object in
ISR context would implicitly be racey and introduce an element of
non-determinism.
Assert that k_event_init() is not called from ISR context.
Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>
The `nocache` is not loadable, thus data stored therein cannot be
initialized by the startup code. This might be needed in special
cases. E.g. One might have a buffer which one wants to DMA into,
and which is a member of a struct. Other members of the struct one
may want to have initialized by the startup code.
The buffer thus should be placed in the `nocache` region, but for
the other members of the buffer to be initialized by the startup
code, the `nocache` region needs to be loadable.
Fix it by making the `nocache` region loadable. Adding a KConfig
symbol to do this optionally was considered, but deemed unnecessary
during the PR.
Signed-off-by: Julian Achatzi <mail@achatzi.pro>
Two issues:
- is_condition_met() was missing proper code for The
K_POLL_TYPE_PIPE_DATA_AVAILABLE case
- z_handle_obj_poll_events() was misplaced in z_impl_k_pipe_write()
Note: I added support for the deprecated pipe implementation to
is_condition_met() but that is untested.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
If something is tagged as nocache it didn't got cleared, which could
lead some weird behaviour where bss memory is non-zero.
Signed-off-by: Peter van der Perk <peter.vanderperk@nxp.com>
Handle both of these sections in a single chunk of code instead of
separately. We don't need to use the legacy .ctors ABI as both
the constructors array and startup logic are managed within a single
link result.
This can now also be used with ARC MWDT which had been using the .ctors
sections but with .init_array semantics. For ARC MWDT, we now always
discard .dtors and .fini sections as Zephyr will never cause global
destructors to execute. Stop discarding .eh_frame sections so that
exception handling works as expected.
When building a NATIVE_APPLICATION, we ask the native C library to run all
of the constructors to ensure any non-Zephyr constructors are run before
main is invoked. It might be "nice" to split the constructors so that the
Zephyr constructors were executed by the Zephyr code while the non-Zephyr
ones were executed by the native C library. I think that could be done if
we knew the pathnames of either the Zephyr or non-Zephyr files. That might
make a good future enhancement.
Signed-off-by: Keith Packard <keithp@keithp.com>
z_tick_sleep function needs to calculate expected wakeup tick. It
required reading current system clock tick. It can be costly since
it is a register access. When timeout is added this value is
calculated and z_add_timeout returns it. It can be used instead to
optimize sleep performance.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Add return value to z_add_timeout. It returns system tick when timeout
will expire.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Accessing system timer registers can be costly and it shall be avoided
if possible. When thread is waken up in z_tick_sleep it may be because
timeout expired or because thread was waken up before sleeping period
passed.
Add function to detect if timeout is aborted (before it was expired).
Use it in the sleep function and avoid reading system ticks if timeout
was not aborted.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Make `k_work_queue_config cfg` as `static const` to enable
compile-time instantiation instead of runtime allocation.
This modification saves substantial flash memory but has system-wide
effects that should be considered.
Signed-off-by: Pisit Sawangvonganan <pisit@ndrsolution.com>
sys_clock_elapsed requires access to system clock register interface
which is often slow. When new relative timeout is added sys_clock_elapsed()
is called once to calculate delta ticks and then if that triggers setting
new timeout sys_clock_elapsed() is called again. This call is redundant
since everything happens under spin lock so it is better to reuse value
returned by the first call.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
When k_malloc() is expressed in terms of k_aligned_alloc() it invokes a
longer aligned allocation code path with an extra runtime overhead even
though no alignment is necessary.
Let's reference and invoke the aligned allocation code path only when an
actual aligned allocation is requested. This opens the possibility for
the linker to garbage-collect the aligning code otherwise.
Also bypass k_heap_malloc() and friends given they're invoked with
K_NO_WAIT. Go directly to sys_heap_*() instead to cut some more unneeded
overhead.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
In .../modules/debug/percepio/TraceRecorder/kernelports/Zephyr/include/\
tracing_tracerecorder.h there is a concealed non-parameterized direct
reference to a local variable that is no longer in scope. Provide a dummy
stub for compilation to succeed until that module's layering violation is
fixed, after which this could be reverted.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
When k_heap_alloc() is expressed in terms of k_heap_aligned_alloc()
it invokes a longer aligned allocation code path with an extra runtime
overhead even though no alignment is necessary.
Let's reference and invoke the aligned allocation code path only when an
actual aligned allocation is requested. This opens the possibility for
the linker to garbage-collect the aligning code otherwise.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Let's avoid division and modulus operations as they're costly... and even
more so when they're unnecessary as in this case the main constraint is
about the alignment being a power of 2 which is a very small subset of
sizeof(void *) multiples.
Then move the assertion to common code for wider coverage.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Convert z_clock_hw_cycles_per_sec to unsigned int to increase
supported frequency range.
Signed-off-by: Anisetti Avinash Krishna <anisetti.avinash.krishna@intel.com>
k_thread_priority_set() on a pended thread wasn't re-inserting into the
waitq, causing the incorrect thread to run based on priority. When using
the scalable waitq config, this can also break assumptions of the tree
and leave the owner of a waitq still being in the waitq tree, cycles in
the tree, or a crash.
Remove and re-add a thread to a waitq to ensure the waitq remains in
order and the tree's assumptions are not violated.
To illustrate the issue, consider 4 threads in decreasing priority
order: A, B, C, and D along with two mutexes, m0 and m1. This is
implemented in the new complex_inversion mutex_api test.
1. D locks m1
2. C locks m0
3. C pends on m1
4. B pends on m1
5. A pends on m0, boosts C's priority, now tree on m1 is not sorted
6. D unlocks m1, left-most thread on tree is B. When removing B from
tree it cannot be found because it searches to the right of C due to
C's boosted priority when the node is actually on the left. rb_remove
silently fails.
7. B unlocks m1, left-most thread on tree is still B and it tries to
unpend itself, resulting in a NULL pointer dereference on
B->base.pended_on.
Signed-off-by: Josh DeWitt <josh.dewitt@garmin.com>
Introduces the Z_IS_TIMEOUT_RELATIVE() macro to help ensure that
checking for relative/absolute timeouts is consistent. Using this
macro also helps ensure that we get the correct behavior when using
32-bit timeouts (CONFIG_TIMEOUT_64BIT=n).
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Improve naming of the scheduler and call it what it is: simple. Using
'dumb' for the default scheduler algorithm in Zephyr is a bad idea.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The current_fp field in the z_kernel structure is only used
by 32-bit x86 (which does not support SMP). As such, it should
reside in the arch specific of section of _kernel.cpus[0].
This also changes the name of 'current_fp' to 'fpu_owner' to
be more consistent with other architectures.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
CONFIG_DYNAMIC_BOOTARGS currently depends on CONFIG_MULTIBOOT_INFO
and CONFIG_BUILD_OUTPUT_EFI which are x86 specific.
This change removes those dependencies so that another architecture
can use CONFIG_DYNAMIC_BOOTARGS without the x86 specifc features.
Signed-off-by: William Tambe <williamt@cadence.com>
Fix compilation with CONFIG_TIMER_READS_ITS_FREQUENCY_AT_RUNTIME=y and
CONFIG_SYS_CLOCK_HW_CYCLES_PER_SEC unset at the board/SoC level.
Signed-off-by: Henrik Brix Andersen <henrik@brixandersen.dk>
Remove restrictions from device_init by allowing to perform device
initialization if the device state flags it being not initialized.
This makes the API usable in contexts where device_deinit has been
called before.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Add a new API to de-initialize a device. When a device is
de-initialized, it will release any resources it has acquired
(e.g. pins, memory, clocks, DMA channels, etc.) and its status
will be left as in its reset state.
It is the responsability of the caller to ensure that the device is
ready to be de-initialized.
For now, deinit call always initializes to NULL. New macros will be
introduced to not break existing device APIs.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Instead of passing a single init function, create
struct device_ops with the init function inside. This allows to easily
extend device's capabilities in the future without too much breakage,
e.g. to add a de-init call.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Introduce a new field to store device flags. Only device deferred init
flag has been added, replacing usage of linker hackery to know wether a
device requires initialization at boot time or not. This change will be
helpful in the near future as devices will become reference counted, so we
will need to know wether they have been initialized or not.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Such union is rather redundant, considering a simple const cast can be
done when initializing the init entry. Note that the init_entry does not
need to be touched now that struct device stores the init call. It is
merely an init entry sorted by linker scripts, so we can intertwine
devices and SYS_INIT.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Device init function is no longer taken from `struct init_entry`, so
there's no need to keep such union.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
It just complicates things. It is not C99 strandard, and since C11 is
not mandatory, it is better to play safe here.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The 'order_key' field in the thread structure '_thread_base' is only
required when CONFIG_SCHED_SCALABLE and/or CONFIG_WAITQ_SCALABLE are
enabled (neither of which is a default setting). Making the existence
of this field conditional slightly reduces the size of the k_thread
structure when neither of those Kconfig options are selected.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
k_sleep() returns a 32 bit count of milliseconds, as that was its
historical API. But it now accepts a potentially 64 bit tick count as
an argument, leading to situations where an early wakeup will produce
sleep times that aren't representable. Clamp this instead of
truncating to an arbitrary value.
Naive code will likely do the right thing with the large return (just
sleeping an extra round), and sophisticated apps can detect INT_MAX to
enable more elaborate retry logic.
(Also fixes a somewhat unfortunate puncutation error in the docs that
implied that it returns zero on early wakeup!)
Fixes: #84669
Signed-off-by: Andy Ross <andyross@google.com>
Currently, the soc/stm32/ccm.ld is not handled in
CMAKE_LINKER_GENERATOR.
This commit adds support, making STM32 supportable by
alternative linkers such as AC6 and IAR.
This commit also renames a variable to match all other
LOADADDR symbols.
Signed-off-by: Robin Kastberg <robin.kastberg@iar.com>
Calling a (handled/ignored) panic with the scheduler lock held
produces spinlock errors in some circumstances, depending on whether
or not the swap gets reached before another context switch. Release
the lock around the call, we don't touch any scheduler state on the
path to z_swap(), so this is safe.
Signed-off-by: Andy Ross <andyross@google.com>
Setting CONFIG_NUM_PREEMPT_PRIORITIES to 128 causes the idle thread to be
assigned priority 128, which exceeds the int8_t range. This results in the
idle thread being assigned the highest priority (-128) instead of the
lowest, causing threads to not wake up from k_sleep.
Restrict the range of CONFIG_NUM_PREEMPT_PRIORITIES to 0 to 127 to ensure
the idle thread always has the lowest priority.
Signed-off-by: Jonas Spinner <jonas.spinner@burkert.com>
With spinlock debugging enabled LLEXTs need additional symbols
exported by the kernel.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
The essential thread check and panic happens at the top of
k_thread_abort(). This is arguably a performance bug: the system is
going to blow up anyway no matter where we put the test, we shouldn't
add instructions to the path taken by systems that DON'T blow up.
But really it's more of a testability/robustness glitch: if you have a
fatal error handler that wants to catch this panic (say, a test using
ztest_set_fault_valid()), then the current code will panic and
early-exit BEFORE THE THREAD IS DEAD. And so it won't actually die,
and will continue on causing mayhem when presumably the handler code
expected it to have been aborted.
It's sort of an unanswerable question as to what the "right" behavior
is here (the system is, after all, supposed to have panicked!). But
this seems preferable for definable practical reasons.
Kill the thread, then panic. Unless it's _current, in which case
panic as late as possible for maximum coverage of the abort path.
Fixes: #84460
Signed-off-by: Andy Ross <andyross@google.com>
K_KERNEL_STACK_RESERVED can be 0 which can give a warning with
-Wtype-limits. Only perform the check if ARCH_KERNEL_STACK_RESERVED
is set. Also remove the the unncessary sets in arch.h where it's
manually set to 0, it defaults to 0 anyways.
Signed-off-by: Ryan McClelland <ryanmcclelland@meta.com>
The routine k_thread_suspend() has a fast path for non-SMP when
suspending the current thread. When SMP is enabled, it is expected
that the compiler drop the entire fast path checks because the whole
expression would always evaluate to false. However, the compiler has
been observed to only drop whole fast path check when the
"!IS_ENABLED(CONFIG_SMP)" condition appears at the beginning of the
fast path check.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
z_reschedule() already has a check to determine if it is called from
the context of an ISR--no need to duplicate it in k_wakeup().
Furthermore, if the target thread is not sleeping, there is no need
to reschedule and we can do a fast return.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
1. Fixes a performance issue in k_msgq_put() to allow for a fast return
path when handling a poll event does nothing.
2. Allows for a fast return path in k_msgq_purge() when no threads were
awakened.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Updates the queue code to both allow for a fast return path in a
few routines when the operation did not wake or signal another
thread.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
When doing a condition variable broadcast, a full reschedule
is only needed if at least one thread was awakened.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Propagates the return value from z_handle_obj_poll_events()
within the message queue, pipes, queue and semaphore objects.
This allows the kernel object code to determine whether it
needs to perform a full reschedule, or if it can perform a
more optimized exit strategy.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
