xTaskRemoveFromUnorderedEventList is a scheduler internal function, thus will
never be called by users. Therefore, the compatbility define is not necessary.
FromISR APIs would put an unblocked task on the pending ready list if
the scheduler is suspended on the current core, irrespective of the
task's core affinity and the state of the scheduler on the other core.
This commit updates this behavior by allowing tasks to get scheduled on
a core which has the scheduler running as long as the task's core
affinity allows it.
The task snapshot tests use esp_cpu_stall() to stall the other CPU before
manually walking the task lists. However, it is possible that the other CPU
was also accessing the task lists when esp_cpu_stall() is called, leading to
flakey tests
This commit fixes the test by using a 2-way handshake instead of
esp_cpu_stall().
The default implementation of some xtensa vectors were previously moved to
"xtensa_vector_defaults.S" as weak functions so that they could be overriden.
This commit moves these default vectors back into "xtensa_vectors.S" in
preparation for further refactoring of the xtensa component.
This commit adds a common macro for the differing configASSERT() calls that
check if the scheduler is suspended outside a critical section. This macro
allows vanilla behavior to be restored for single core builds. Also added
comments explaining why xTaskGetSchedulerState() is called instead when
building for SMP.
Vanilla FreeRTOS expects applications to use one of the heap implementations
provided by FreeRTOS (i.e., heap_x.c), where functions such as pvPortMalloc()
and vPortFree() are defined in the heap implementation.
However, ESP-IDF already provides its own heap implementation
(i.e., esp_heap_caps.h). Thus, the pvPortMallc()/vPortFree() functions were
previously overriden by macro to call esp_heap functions directly.
This commit refactors the FreeRTOS port's heap as such:
- Added a heap_idf.c that implements all of the heap related functions required
by FreeRTOS source
- All dynamic memory allocated by FreeRTOS is from internal memory. Thus, the
FreeRTOS heap is the internal memory subset of the ESP-IDF heap.
- Removed some old macros to reduce diff from upstream source code.
The stream buffer send-receive test preivously contained the following bugs:
- "sender" task would send 101 bytes instead of 100
- The main task would return before "sender" task sends its 101st item. Thus
the sender task would cause memory corruption when accessing "tc" structure
allocated on the main task's stack
This commit fixes and simplifies the stream buffer test.
This commit updates the linker fragment file for IDF FreeRTOS.
- Linker fragment file's formatting was updated
- Placement rules of functions is now clearly specified inside the linker
fragment file.
- Some extra functions are now placed in flash in accordance to the new
placement rules.
This commit updates the linker fragment file for Amazon SMP FreeRTOS.
- Linker fragment file's formatting was updated
- Placement rules of functions is now clearly specified inside the linker
fragment file.
- Some extra functions are now placed in flash in accordance to the new
placement rules.
This commit adds a separate linker fragment file "linker_common.lf" for the
functions in "port_common.c". The placement rules are now clearly specified
inside the linker fragment file.
This commit updates the FreeRTOS port FPU unit tests so that they run for
multiple iterations, thus checking that a task's FPU context is properly
cleaned up on deletion.
Previously, IDF FreeRTOS would restrict the clean up of task memory (done by
vTaskDelete() or the Idle task) to only tasks pinned to the current core or
unpinned tasks. This was due to the need to clear the task's coprocessor
ownership on the other core (i.e., "_xt_coproc_owner_sa"). But this restriction
can be lifted by simply protecting access of "_xt_coproc_owner_sa" with a
spinlock.
This commit implements a "_xt_coproc_owner_sa_lock" to protect the access of
"_xt_coproc_owner_sa", thus vTaskDelete() and prvDeleteTCB() can now delete
tasks pinned to the other core so long as that task is not currently running.
Note: This fix was copied from the Xtensa port of Amazon SMP FreeRTOS
This commit removes the dependency on portUSING_MPU_WRAPPERS on the Xtensa port
of IDF FreeRTOS. This dependency was added due to a hack implemented in the
upstream port that required the usage of the "xMPUSettings" member of the TCB.
The "xMPUSettings" would be used as a pointer to the task's coprocessor save
area on the stack, even though FreeRTOS MPU support was not available.
The hack has now been removed, and the CPSA pointer is now calculated using
a combination of constant offsets values and the pxEndOfStack member of the
TCB.
Note: This impelemtation was copied from the Xtensa port of Amazon SMP FreeRTOS.
This commit refactors the FPU handling code on the Xtensa port of Amazon SMP
FreeRTOS in the following ways:
Auto-pinning via XT_RTOS_CP_EXC_HOOK
------------------------------------
The "_xt_coproc_exc" exception would previously automatically pin a task that
uses the FPU to the current core (to ensure that we can lazy save the task's FPU
context). However, this would mean that "xtensa_vectors.S" would need to be
OS-aware (to read the task's TCB structure).
This is now refactored so that "_xt_coproc_exc" calls a CP exception hook
function ("XT_RTOS_CP_EXC_HOOK") implemented in "portasm.S", thus allowing
"xtensa_vectors.S" to remain OS agnostic.
Using macros to acquire owner spinlock
--------------------------------------
The taking and relasing of the "_xt_coproc_owner_sa_lock" is now mostly
abstracted as the "spinlock_take" and "spinlock_release" macro. As a result,
"_xt_coproc_release" and "_xt_coproc_exc" are refactored so that:
- They are closer to their upstream (original) versions
- The spinlock is only taken when building for multicore
- The spinlock held region is shortened (now only protects the instructions
that access the "_xt_coproc_owner_sa" array
Other Changes
-------------
- Updated placing and comments of various "offset_..." constants used by
portasm.S
- Update description of "get_cpsa_from_tcb" assembly macro
- Tidied up some typos in the ".S" files
This commit fixes the following FPU/Coprocessor bugs in the Xtensa port of
Amazon SMP FreeRTOS:
- vPortCleanUpCoprocArea() does not calculate the correct pointer to the
task's CPSA (located on the task's stack). This can result in
- _xt_coproc_release() not releasing the task's CP ownership
- The next coprocessor exception will write the current CP owner (i.e., the
deleted task's CPSA) leading to memory corruption
- _xt_coproc_release() writes xCoreID instead of 0 when clearing a CP owner.
This results in the next CP exception trying to load the CP owner's CPSA at
the address of "xCoreID", leading to a double exception.
This commit updates variousf pytest scripts to expect
"main_task: Calling app_main()" instead of "cpu_start: Starting scheduler" as
indicator of the start of an application.
This commit refactors port_common.c so that it only contains implementation of
FreeRTOS port functions that are common to all FreeRTOS ports (i.e., on all
architectures and on all FreeRTOS implementations).
This commit refactors the OS startup functions as follows:
- Moved the OS/app startup functions listed below to "app_startup.c". Their
implementations are now common to all ports (RISC-V and Xtensa) of all
FreeRTOS implementations (IDF and Amazon SMP).
- esp_startup_start_app()
- esp_startup_start_app_other_cores()
- Removed esp_startup_start_app_common() as app startup functions are now
already common to all ports.
- Added extra logs to "main_task" to help with user debugging
Note: Increased startup delay on "unity_task". The "unity_run_menu()" is non
blocking, thus if the main task or other startup tasks have not been freed
by the time "unity_run_menu()" is run, those tasks will be freed the next time
"unity_task" blocks. This could cause some tests to have a memory leak, thus
the "unity_task" startup delay has increased.
FreeRTOS(IDF): Restore vanilla task selection algorithm if building for single core with optimized selection is disabled
See merge request espressif/esp-idf!21001
Since core affinity is supported in task stats by default in Amazon
FreeRTOS SMP, the Kconfig option
CONFIG_FREERTOS_VTASKLIST_INCLUDE_COREID is now made only available for
IDF-FreeRTOS.
This commit restores the vanilla behavior for taskSELECT_HIGHEST_PRIORITY_TASK
when configNUM_CORES == 1 && configUSE_PORT_OPTIMISED_TASK_SELECTION == 0. This
results in the SMP selection algorithm (taskSelectHighestPriorityTaskSMP) to
only be run when configNUM_CORES > 1.
This commit adds preprocessing time checks to ensure that
- configNUM_CORES is defined to 1 or 2
- Incompatible configurations are not set depending on configNUM_CORES
Macros that need to reteurn a value should use GCC statement expression macro
syntax. This commit fixes the portTRY_ENTER_CRITICAL() in the RISC-V port of
SMP FreeRTOS to be a statement expression macro.
Due to SMP, critical sections have been added to xTaskIncrementTick() and
vTaskSwitchContext() (to take the xKernelLock). However, this is technically
not necessary when building for single-core as FreeRTOS expect these funcitons
to be called with interrupts already disabled.
This commit makes the critical secitons in those functions depend on
"configNUM_CORES > 1", and ensures that interrupts are disabled when calling
those functions. This effectively restores the vanilla behavior for these
functions when building for single-core.
In IDF FreeRTOS, when building for SMP, there are numerous functions
which require different critical sections when compared to single-core. This
commit encapsulates those difference into a common set of macros whose
behavior depends on "configNUM_CORES > 1". As such...
- Vanilla behavior has been restored for some functions when building for
single core (i.e., used to call taskENTER_CRITICAL, now disables interrupts
mactching vanilla behavior).
- Reduces number of "#ifdef (configNUM_CORES > 1)" in functions
- Any SMP only critical sections are now wrapped by
"#ifdef (configNUM_CORES > 1)" and properly documented via comments.
The portDISABLE_INTERRUPTS() macro on Xtensa should return only the interrupt
mask/level before the interrupts were disabled. Previously, the entire contents
of PS register were returned (i.e., direct return from RSIL instruction without
any bit masking or shifting).
This commit fixes the portDISABLE_INTERRUPTS() macro to return the INTLEVEL
bitfield of the PS register.
This commit updates how configTHREAD_LOCAL_STORAGE_DELETE_CALLBACKS is defined
by default it to 0 if not defined elsewhere. Dependent code now check for
"configTHREAD_LOCAL_STORAGE_DELETE_CALLBACKS == 1" instead.
This commit synchronize the cosmetic differences of IDF FreeRTOS to upstream
Vanilla v10.4.3. Comsetic differences include:
- Out of date doxygen API descriptions
- Misnamed parameters
- Missing examples
- Fixed/added missing @cond/@code directives
- Extra/missing comments/lines
- Code formatting (uncrustify)
Other changes:
- Some ESP_PLATFORM directives were also removed
- xTaskIncrementTickOtherCores() now depends on "configNUM_CORES > 1"
- Updated some multi-core dummy variable names in FreeRTOS.h
This commit removes the updates the usage of queue locks in IDF FreeRTOS
Queue locks are present in Vanilla FreeRTOS to ensure that queue functions
behave deterministicly in critical sections (i.e., no walking linked lists
while interrupts are disabled). However, currently in IDF FreeRTOS...
- When configNUM_CORES > 1, IDF FreeRTOS drops the determinism requirement.
Thus, queue functions could be simplified if queue locks were not used at all
(and have a queue function do everything inside the same critical section).
- When configNUM_CORES == 1, the current queue implementation in IDF FreeRTOS
does not meet the determinism requirements, as critical sections are used
(instead of scheduler suspension) when locking/unlocking the queues.
There, this commit updates multiple queue functions so that
- When configNUM_CORES > 1
- Queue locks are no longer used. All actions are done within the same
critical section.
- Affected queue functions now need 40% less CPU clock cycles when blocking
- When configNUM_CORES = 1
- Queue locks are still used.
- Vanilla behavior of suspending the scheduler is restored when locking
the queue. Thus queue fucntions are now deterministic and have the same
behavior as Vanilla FreeRTOS.
- Affected queue functions now takes 36% more CPU clock cycles when
blocking (due to the scheduler suspension/resumption).
The following changes have been made in this commit:
1. configMINIMAL_STACK_SIZE is now defined as CONFIG_FREERTOS_IDLE_TASK_STACKSIZE.
2. Removed configIDLE_TASK_STACK_SIZE config as it was redundant.
3. Updates the order of allocating the TCB and stack memory to avoid the
stack memory overriding the TCB memory when the stack grows downwards.
4. CONFIG_FREERTOS_IDLE_TASK_STACKSIZE is now incorporated into the
FreeRTOSConfig_smp.h to configure the IDLE0 stack size.
This commit fixes flakey task delay tests when running on IDF FreeRTOS. IN IDF
FreeRTOS, core 1's tick interrupts could be up to (< 1 tick) out of phase from
core 0's tick interrupt, thus the measured delay time on core 1 could last 1
tick longer. This commit makes the test detla 2 ticks for IDF FreeRTOS.
This commit adds a copy of upstream vanilla FreeRTOS uncrustify.cfg files from
the following upstream commit:
89e4823a49
Note: "pp_indent_brace" is commented out as that was enabled after v10.4.3
Also also added a note in the ESP-IDF style guide aobut formatting FreeRTOS
source code using Uncrustify.
This commit refactors the way stream buffers initialize their spinlock.
- "prvInitialiseNewStreamBuffer()" now initializes the stream buffer fields
manually (instead of using memset()) to avoid resetting the spin lock
- Stream buffer creation functions now manually initialize the spinlock after
the other fields are initialized using "prvInitialiseNewStreamBuffer()"
Also added comments to event group spinlock initializtion.
The llvm asm pasrser doesn't support some gnu asm extensions,
like using "&" with macro arguments. So, replace such code with
code which could be compiled by llvm.
An explicit check of the mutex owner during a mutex release operation is
not necessary for FreeRTOS SMP as this is checked by default. Hence, the
corresponding Kconfig option is removed.
This commit fixes/ignores flakey freertos unit tests after migrating them to
the test app:
- Added vTaskDelay() before teardown to prevent memory leaks
- Adjusted the "main" task's priority so that scheudling tasks would work
- pytest now only runs tests that are not ignored
- Reset tests are temporarily ignored. Will be enabled to dedicate reset tests.
- Some flakey tests are fixed by adjusting delays and stack sizes.
This commit adds the sdkconfig files for the FreeRTOS test app. These
configurations were dervied from the various legacy unit test app's
config files that included the FreeRTOS component.
This commit tries to keep a 1 to 1 config parity with the legacy test app.
Meaning, if FreeRTOS test were run on a particular target with a particular
config, that config will be represented in one of the test app's
sdkconfig.ci.XXX files.
However, the following configurations were removed for FreeRTOS tests:
- The "freertos_flash" option was removed due to redudancy (already tested in
freertos_options)
Some tests were placed in the incorrect test groups (i.e., kernel, port,
performance etc). This commit fixes those placements.
The following redundant tests were also removed:
- "test_panic.c" as behavior is already covered in esp_system tests
This commit fixes a bug where if an unpinned task is interrupted by a level 1
ISR that users the FPU, the FPU usage will cause the interrupted task to
become pinned to the current core.
Note: This bug was already fixed in SMP FreeRTOS in commit
d69361779e. This commit simply backports the
fix to IDF FreeRTOS.
This commit refactors the existing FPU unit tests as follows
- Rename them from coproc to FPU
- Reorganize test placement
- Make existing tests work on both IDF and SMP FreeRTOS
- Update test documentation
- Remove old "test_float_in_isr.c"
This commit refactors the pxPortInitialiseStack() function of the riscv
FreeRTOS ports (both IDF and SMP FreeRTOS).
- Each stack area is now separated into their own functions
- Each function will individually
- Push the stack pointer to allocate the stack area
- Initiaze the allocated stack area
- Each stack area's size and usage is now clearly documented in code
This commit refactors the pxPortInitialiseStack() function of the xtensa
FreeRTOS ports (both IDF and SMP FreeRTOS).
- Each stack area is now separated into their own functions
- Each function will individually
- Push the stack pointer to allocate the stack area
- Initiaze the allocated stack area
- Each stack area's size and usage is now clearly documented in code
* Added port layer from the FreeRTOS POSIX port, added
additional port code for ESP-IDF.
* Created another hello world example using that POSIX
port in tools/test_apps.
* Removed old linux app
This commit adds a priority scheduling unit test to test that the scheduler
always schedules the highest priority ready state task. Single core and SMP
variants of the test are both added.
This commit adds a README.md containing guidelines on how to refactor the
FreeRTOS unit tests for upstreaming. A portTestMacro.h header was also added
which contains port implementation specific macros used by the test cases.
This commit places some functions in port.c and port_common.c into flash by
default. These functions are mostly called once during startup, thus shouldn't
affect applications much when placed into flash instead of the default IRAM.
This commit synchronizes multiple functions in tasks.c with upstream v10.4.3
that contain minor differences.
The following functions have had their parameter names or code formatting
updated:
- xTaskCreateStaticPinnedToCore()
- xTaskCreatePinnedToCore()
- prvInitialiseNewTask()
- prvTaskIsTaskSuspended()
- vTaskStartScheduler()
- xTaskResumeAll()
- xTaskCheckForTimeOut()
- uxTaskResetEventItemValue()
The following functions have had their missing "xAlreadyYielded" restored
- xTaskDelayUntil()
- vTaskDelay()
The following functions have had their critical section/interrupt disdable
usage update
- vTaskSuspendAll()
- xTaskGetTickCountFromISR()
- xTaskGetApplicationTaskTagFromISR()
This commit synchronizes multiple functions that contain major differences with
upstream v10.4.3. Multi-core modifications are then reapplied to the function.
Vanilla code uses "pxTCB == pxCurrentTCB" to check whether a task is currently
running. This commit adds the following macros instead.
- taskIS_CURRENTLY_RUNNING()
- taskIS_CURRENTLY_RUNNING_ON_CORE()
The following functions have been updated to use those macros instead.
- eTaskGetState()
- vTaskPrioritySet()
- vTaskResume()
The following functions have been synchronized with upstream v10.4.3.
Multi-core modifications are then reapplied.
- vTaskSuspend()
- vTaskDelete()
- prvGetExpectedIdleTime()
This commit synchronizes multiple functions in queue.c with upstream v10.4.3.
Multi-core modifications are then reapplied to these functions. The following
functions were modified:
prvNotifyQueueSetContainer()
xQueueGenericCreateStatic()
xQueueGenericCreate()
xQueueGetMutexHolder()
xQueueCreateCountingSemaphoreStatic()
xQueueCreateCountingSemaphore()
xQueueGenericSend()
xQueueGenericSendFromISR()
xQueueReceiveFromISR()
uxQueueMessagesWaiting()
prvUnlockQueue()
prvIsQueueFull()
xQueueAddToSet()
xQueueRemoveFromSet()
prvNotifyQueueSetContainer()
Note: The SEGGER_SYSVIEW traceQUEUE_SEND() macro was updated as the
xCopyPosition argument is no longer available in scenarios where the
macro is called.
This commit fixes the function declaration naming from esprv_intc_set_threshold()
to esprv_intc_int_set_threshold(), thus allowing the underlying ROM funciton to be
exposed via the header.
This commit removes the riscv_interrupts.h header is it has become redundant. The previously
exposed API has been handled as follows:
- "riscv_interrupt_enable()" and "riscv_interrupt_disable()" have been removed. These functions
were declarations only and never had any implementation.
- "riscv_global_interrupts_enable()" and "riscv_global_interrupts_disable()" renamed to
"rv_utils_intr_global_enable()" and "rv_utils_intr_global_disable()" respectively and now
placed in rv_utils.h
ESP32-C2 has a single group timer, thus it will use it for the interrupt watchdog,
which is more critical than the task watchdog. The latter is implement in
software thanks to the `esp_timer`component.
This commit synchronizes multiple functions in tasks.c with upstream v10.4.3
that contain minor differences.
The following functions have had their parameter names or code formatting
updated:
- xTaskCreateStaticPinnedToCore()
- xTaskCreatePinnedToCore()
- prvInitialiseNewTask()
- prvTaskIsTaskSuspended()
- vTaskStartScheduler()
- xTaskResumeAll()
- xTaskCheckForTimeOut()
- uxTaskResetEventItemValue()
The following functions have had their missing "xAlreadyYielded" restored
- xTaskDelayUntil()
- vTaskDelay()
The following functions have had their critical section/interrupt disdable
usage update
- vTaskSuspendAll()
- xTaskGetTickCountFromISR()
- xTaskGetApplicationTaskTagFromISR()
This commit synchronizes multiple functions that contain major differences with
upstream v10.4.3. Multi-core modifications are then reapplied to the function.
Vanilla code uses "pxTCB == pxCurrentTCB" to check whether a task is currently
running. This commit adds the following macros instead.
- taskIS_CURRENTLY_RUNNING()
- taskIS_CURRENTLY_RUNNING_ON_CORE()
The following functions have been updated to use those macros instead.
- eTaskGetState()
- vTaskPrioritySet()
- vTaskResume()
The following functions have been synchronized with upstream v10.4.3.
Multi-core modifications are then reapplied.
- vTaskSuspend()
- vTaskDelete()
- prvGetExpectedIdleTime()
This commit synchronizes multiple functions in queue.c with upstream v10.4.3.
Multi-core modifications are then reapplied to these functions. The following
functions were modified:
prvNotifyQueueSetContainer()
xQueueGenericCreateStatic()
xQueueGenericCreate()
xQueueGetMutexHolder()
xQueueCreateCountingSemaphoreStatic()
xQueueCreateCountingSemaphore()
xQueueGenericSend()
xQueueGenericSendFromISR()
xQueueReceiveFromISR()
uxQueueMessagesWaiting()
prvUnlockQueue()
prvIsQueueFull()
xQueueAddToSet()
xQueueRemoveFromSet()
prvNotifyQueueSetContainer()
Note: The SEGGER_SYSVIEW traceQUEUE_SEND() macro was updated as the
xCopyPosition argument is no longer available in scenarios where the
macro is called.
This commit synchronizes the following xTaskRemoveFrom...EventList() with
upstream v10.4.3. Multi-core modifications are then reapplied to the
functions.
The functions updated are:
- xTaskRemoveFromEventList()
- vTaskRemoveFromUnorderedEventList()
The following multi-core macros have been removed as they are no longer
required.
- tskCAN_RUN_HERE()
- taskYIELD_OTHER_CORE()
This commit synchronizes the following vTaskPlaceOnEventList...() with
upstream v10.4.3.
The functions updated are:
- vTaskPlaceOnEventList()
- vTaskPlaceOnUnorderedEventList()
- vTaskPlaceOnEventListRestricted()
The traceTASK_DELAY_UNTIL() macro has also been updated.
This commit reverts the previous "taskENTER_CRTIICAL();" so that the argument
is now provided for better code readability. The names of the spinlocks have
also been updated.
This commit updates various FreeRTOS functions to call the newly added
prvCheckForYield() and prvCheckForYieldUsingPriority() when checking for
yielding. This allows the source code to match upstream more closely.
When a FreeRTOS function unblocks a task, that function will check whether
the unblocked task requires a yield to be called. This is currently done by
having each function individually check if the unblocked task has a higher
priority than the both cores, and yielding the appropriate core.
This commit adds the macros list below to abstract away the yielding checking
procedure. This will allow the code to match upstream more closely.
- prvCheckForYield()
- prvCheckForYieldUsingPriority()
The event group unit tests would previously use a single call bit (the
"BIT_CALL" macro) to unblock all the test tasks. However, if one or more tasks
were delayed in calling xEventGroupWaitBits(), the other tasks would clear the
test bits, leading to the test deadlocking.
This commit updates the bits used so that each task gets their own CALL and
RESPONSE bits.
For RISC-V and Xtensa targets, in case a panic needs to happen when
Task WDT is triggered (ESP_TASK_WDT_PANIC), the interruptee's stack
is now used for printing the backtrace.
Abort after Task Watchdog is triggered can happen on APP CPU (second core).
Closes https://github.com/espressif/esp-idf/issues/7207
The description of how the xTaskGenericNotifyWait parameter is handled in the
xTaskGenericNotifyWait() function was inaccurate.
In this commit, the description was updated to match the implementation of xTaskGenericNotifyWait().
This commit marks all functions in interrupt_controller_hal.h, cpu_ll.h and cpu_hal.h as deprecated.
Users should use functions from esp_cpu.h instead.
This function removes the following legacy atomic CAS functions:
From compare_set.h (file removed):
- compare_and_set_native()
- compare_and_set_extram()
From portmacro.h
- uxPortCompareSet()
- uxPortCompareSetExtram()
Users should call esp_cpu_compare_and_set() instead as this function hides the details
of atomic CAS on internal and external RAM addresses.
Due to the removal of compare_set.h, some missing header includes are also fixed in this commit.
This commit fixes vTaskSuspendAll() and xTaskResumeAll() in the following ways.
- For vTaskSuspendAll()
- Sync function source code with upstream single core version
- Clearly mark IDF additions
- For xTaskResumeAll()
- Sync function source code with upstream single core version
- Clearly mark IDF additions
- Fix bug where cores other than core 0 were allowed to unwind pended ticks
This commit fixes the following issues:
- Clearly mark IDF additions in vTaskStepTick() and xTaskCatchUpTicks()
- Add back vTaskSuspendAll()/xTaskResumeAll() to xTaskCatchUpTicks(). Scheduler
suspension and resumption is required to unwind the pended ticks.
Previously, xTaskIncrementTick() would always trigger a yield (i.e., return pdTRUE)
when called on core 1. This would mean core 1 would call vTaskSwitchContext() on
every tick, leading to uneccesary CPU overhead.
This commit fixes xTaskIncrementTick() in the following ways:
- Clearly mark IDF additions in xTaskIncrementTick()
- Moved esp_vApplicationTickHook() to be called direclty in xPortSysTickHandler() in order to
simplify xTaskIncrementTick().
- Only core 0 calls xTaskIncrementTick() in order to simplify the function's logic. Core 0 is
solely responsible for managing xTickCount and xPendedTicks
- All other cores call xTaskIncrementTickOtherCores() instead which is a simplified version of
xTaskIncrementTick() and handles the following:
- Check if core 0 has unblocked a higher priority task to run
- Check if the current core requires time slicing
- Call vApplicationTickHook()
This commit fixes the following:
- Fixes a bug wherein the timer status is active for oneshot timers that
expire before they are started.
- Callback calls for periodic timers where made before the timers were
auto-reloaded.
- Timer expiry trace was being set after the timer callback is called.
Closes https://github.com/espressif/esp-idf/issues/8014
Idle tick hook that indicates completion of scheduler init must be launched
on other core than the one where main task runs. Earlier it was assumed that
main task shall always run on PRO cpu but that behavior can be changed with
ESP_MAIN_TASK_AFFINITY_CPU1 and hence this fix.
Closes https://github.com/espressif/esp-idf/issues/9247
Closes IDFGH-7705
FreeRTOS uses a single "xSchedulerRunning" variable to tack whether the
scheduler has started, and this variable is set to "pdTRUE" by core 0
via calling vTaskStartScheduler().
However, with SMP FreeRTOS, there is a race condition where core 0 has
already started the scheduler and another core has not called xPortStartScheduler()
yet and calls some FreeRTOS API. Thus the resultant FreeRTOS API can
cause errors as it thinks the scheduler has started.
This commit adds a temporary workaround (by having each core maintain their
own "xSchedulerRunning" variable.
The "FreeRTOS Event Groups" main task will only wait a single tick for the created
tasks to set their response bits. This short delay may not be sufficent if the tick
frequency is high.
This commit updates the test so that
- the main task waits indefinitely for all the response bits to be set.
- created tasks are cleand up by the main task
FreeRTOS synchronization primitives (e.g., queues, eventgroups) use various event lists (i.e., task lists) to track what
tasks are blocked on a current primitive. Usually these event lists are accessed via one of the event lists functions
(such as vTask[PlaceOn|RemoveFrom]UnorderedEventList()), which in turn ensure that the global task list spinlock
(xTaskQueueMutex) is taken when accessing these lists.
However, some functions in event_groups.c manually traverse their event lists. Thus if a tick interrupt occurs on
another core during traversal and that tick interrupt unblocks a task on the event list being traversed, the event list
will be corrupted.
This commit modifies the following event_groups.c functions so that they take the global task list lock before
traversing their event list.
- xEventGroupSetBits()
- vEventGroupDelete()
The CONFIG_FREERTOS_PLACE_FUNCTIONS_INTO_FLASH allows some FreeRTOS functions to be placed
into flash, thus saving some IRAM. However, this feature was previously not supported for
SMP FreeRTOS.
This commit adds CONFIG_FREERTOS_PLACE_FUNCTIONS_INTO_FLASH support for SMP FreeRTOS. There
are now separate linker fragment files for IDF FreeRTOS and SMP FreeRTOS, named "linker.lf"
and "linker_smp.lf" respectively.
The TLSP deletion callback feature is not compatible with the CONFIG_FREERTOS_ENABLE_STATIC_TASK_CLEAN_UP
option. However, the "freertos_options" unit test configuration will enable that option.
This commit disables all CONFIG_FREERTOS_ENABLE_STATIC_TASK_CLEAN_UP related features when compiling with
SMP FreeRTOS.
prvYieldCore() leads to an array-out-of-bounds error when compiled with
-Os optimization and configNUM_CORES = 1. This commit avoids this
compile warning by compiling out the part of code which is unnecessary
when configNUM_CORES is 1.
The Xtensa FreeRTOS port does not save the threadptr register when
doing a voluntary yield. This can result in a crash when multiple
tasks used the threadptr register and call "taskYIELD()".
This commit adds the threadptr register to the solicited stack frame.
