If a user configures the configASSERT macro to expand to nothing,
a call to the Win32 API SetWaitableTimer() doesn't get compiled.
This can happen if, for example, configASSERT(x) set defined as
assert(x) (from assert.h) which expands to nothing when NDEBUG is
set (common for "release" builds).
This commit does the following:
* Remove xEventGroupClearBitsFromISR and
xEventGroupSetBitsFromISR functions direct mapping
to xTimerPendFunctionCallFromISR in case the trace
facility is feature not set. This should match the
current implementation in which the application calling
xEventGroupClearBitsFromISR
without defining the trace macros
(i.e. traceENTER_xEventGroupClearBitsFromISR is
a blank function in this case) will end up calling
xTimerPendFunctionCallFromISR function directly.
This enhances the readability of the code by not
guarding the xEventGroup<Set/Clear>BitsFromISR
functions' declaration and definition based on
the trace facility macro (i.e. configUSE_TRACE_FACILITY).
* Refactor the guarding macros for MPU_xEventGroupClearBitsFromISR
and MPU_xEventGroupSetBitsFromISR functions to match the
change above.
The current implementation leads to redefinition warnings
as `event_groups.h`redefines xEventGroupClearBitsFromISR
and xEventGroupSetBitsFromISR based on
`configUSE_TRACE_FACILITY' after they have been defined based
on configUSE_MPU_WRAPPERS_V1 in `mpu_wrappers.h`. The
implemented changes should resolve these warnings.
Signed-off-by: Ahmed Ismail <Ahmed.Ismail@arm.com>
Add GCC_ARM_CRX_MPU to the set of FREERTOS_PORT patterns that trigger
adding mpu_wrappers.c and mpu_wrappers_v2.c to freertos_kernel_port
target_sources.
* [MSVC-MingW] Move windows headers to port.c
This prevents the inclusion of windows.h. into all header files using
FreeRTOS.h and thus defining several macros conflicting with common
definitions.
* [MSVC-MingW] Include correct header for compiler intrinsics
---------
Co-authored-by: Gaurav-Aggarwal-AWS <33462878+aggarg@users.noreply.github.com>
GNU as makes unrecognized sections loadable and writable by default, but
LLVM's assembler requires specifying flags explicitly. Without them, the
linker generates "has non-ABS relocation" errors when trying to link the
resulting object files.
Previously, the shared thread key was deleted in xPortStartScheduler
after scheduler was ended. This created a race condition where
prvThreadKeyDestructor (responsible for freeing thread-specific heap
memory) would not be called if xPortStartScheduler deleted the key
before the last task deletion, as destructors are not invoked after key
deletion (see https://github.com/walac/glibc/blob/master/nptl/pthread_create.c#L145-L150).
Move thread key deletion to process exit to ensure all thread-specific
memory is properly freed.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
- In list.h fix a typo in comments.
- In portable/ThirdParty/GCC/Posix/port.c fix compiler warning about
"gcc -Wwrite-strings".
Signed-off-by: Florian La Roche <Florian.LaRoche@gmail.com>
Co-authored-by: Gaurav-Aggarwal-AWS <33462878+aggarg@users.noreply.github.com>
* Fix race in POSIX port `vPortEndScheduler`
The `vPortEndScheduler` checks whether it's a FreeRTOS thread after signalling the scheduler thread to stop. This creates a race between the check and the destruction of the thread key. By moving the signal to the scheduler thread after the check, the race is prevented.
* Code review suggestions
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
---------
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
Co-authored-by: Gaurav Aggarwal <aggarg@amazon.com>
FreeRTOS MPU: Remove MPU region number check
This change removes the assertion and runtime check that enforces
matching between configTOTAL_MPU_REGIONS and physical MPU
regions,. This allows applications running on devices with 16 MPU
regions to manage 8 MPU regions while leaving the remaining 8
for the kernel.
Signed-off-by: Erick Reyes <erickreyes@google.com>
Use architecture-dependent UBaseType_t instead of fixed type for
ullMachineTimerCompareRegisterBase. This type is defined to uint32_t or
uint64_t based on XLEN, resolving warnings on 32-bit platforms.
Reported by landretk@ on the PR FreeRTOS/FreeRTOS-Kernel#1176.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
RISC-V: refine fpu reg context offset
pxCode and mstatus stored at index 0 and 1 are based on XLEN.
Therefore, the correct formula to calculate the FPU register index
should be ( ( 2 * portWORD_SIZE ) + ( regIndex * portFPU_REG_SIZE ) ).
Signed-off-by: wangfei_chen <wangfei_chen@realsil.com.cn>
* port: riscv: Split the number of registers and the size of the context
* port: riscv: Create some macros for the FPU context
* port: riscv: Add a couple of macros that store fpu context
* port: riscv: Update the stack init function to include the fpu context size
* port: riscv: Add a chip_specific_extensions file that includes the F extension
* Update dictionary to include some risc-v instructions
* port: riscv: Fix a few typos
* port: riscv: Apply @aggarg's sugestions
Disable stack overflow check for MPU ports
Stack overflow check is not straight forward to implement for MPU ports
because of the following reasons:
1. The context is stroed in TCB and as a result, pxTopOfStack member
points to the context location in TCB.
2. System calls are executed on a separate privileged only stack.
It is still okay because an MPU region is used to protect task stack
which means task stack overflow will trigger an MPU fault.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
While using the macOS default LLDB debugger, a call to vTaskEndScheduler results in an unhandled SIGUSR1 (aka SIGRESUME) when restoring the scheduler thread's signals with pthread_sigmask. This crashes the program.
Added instructions in portable/ThirdParty/GCC/Posix/port.c to suppress SIGUSR1 to prevent LLDB debugger interference when exiting xPortStartScheduler
Thanks to: @johnboiles for pointing it out in #1224
On FreeBSD pthread_once_t is a struct and cast is required.
Otherwise there's compilation error:
../../mocks/freertos/port.c:261:23: error: expected expression
hSigSetupThread = PTHREAD_ONCE_INIT;
^
PTHREAD_ONCE_INIT is defined as: { PTHREAD_NEEDS_INIT, NULL } on FreeBSD
Co-authored-by: Jakub Tymejczyk <jakub.tymejczyk@enigma.com.pl>
Mark mutex as robust to prevent deadlocks
Prevent application hangs that occur when a thread dies while holding a
mutex, particularly during vTaskEndScheduler or exit calls. This is
achieved by setting the PTHREAD_MUTEX_ROBUST attribute on the mutex.
Fixes:
- GitHub issue: FreeRTOS/FreeRTOS-Kernel#1217
- Forum thread: freertos.org/t/22287
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
Fix the context array size for MPU ports
Ensure the saved context location falls within the reserved context area
rather than overlapping with the next MPU_SETTINGS structure member.
This never caused a problem because actual read/write operations
start from one word before the saved context location.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
The application writer needs to name their IRQ handler as:
1. vApplicationIRQHandler if the IRQ handler does not use FPU registers.
2. vApplicationFPUSafeIRQHandler is the IRQ handler uses FPU registers.
When the application uses vApplicationFPUSafeIRQHandler, a default
implementation of vApplicationIRQHandler is used which stores FPU
registers and then calls vApplicationFPUSafeIRQHandler.
Note that recent versions of GCC may use FP/SIMD registers to optimize 16-bytes
copy and especially when using va_start()/va_arg() functions (e.g printing some thing
in IRQ handlers may trigger usage of FPU registers)
This implementation is heavily inspired by both the ARM_CA9 port and the ARM_CRx_No_GIC
port done in [1]
[1] https://github.com/FreeRTOS/FreeRTOS-Kernel/pull/1113
Signed-off-by: Marouen Ghodhbane <marouen.ghodhbane@nxp.com>
FPSR and FPCR are two 64-bits registers where only the lower 32 bits are defined.
Save them when doing context switch with FPU context saving enabled.
Signed-off-by: Marouen Ghodhbane <marouen.ghodhbane@nxp.com>
This is a direct backport of upstream commit [1] for aarch64 (legacy operation port)
done under [2]
The same code can be applied on the aarch SRE port to be able to enable FPU context
saving on all tasks context switch to mitigate GCC optimization to use SIMD registers
for copy.
[1] "55eceb22: Add configUSE_TASK_FPU_SUPPORT to AARCH64 port (#1048)"
[2] https://github.com/FreeRTOS/FreeRTOS-Kernel/pull/1048
Signed-off-by: Marouen Ghodhbane <marouen.ghodhbane@nxp.com>
Add xQueueCreateSetStatic method for static allocation of Queue Sets
This commit introduces the xQueueCreateSetStatic function, which allows for the static allocation of Queue Sets in FreeRTOS when both configUSE_QUEUE_SETS and configSUPPORT_STATIC_ALLOCATION are enabled.
armv8.1-m: Add task dedicated PAC key
To harden the security, each task is assigned a dedicated PAC key, so that attackers needs to guess the all the tasks' PAC keys right to exploit the system using Return Oriented Programming.
The kernel is now updated to support the following:
* A PAC key set with a random number generated and is saved in the task's context when a task is created.
* As part of scheduling, the task's PAC key is stored/restored to/from the task's context when a task is unscheduled/scheduled from/to run.
stack-overflow-check: Introduce portGET_CURRENT_TOP_OF_STACK macro
When MPU wrapper v2 is used, the task's context is stored in TCB and `pxTopOfStack`` member of TCB points to the context location in TCB. We, therefore, need to read PSP to find the task's current top of stack.
Signed-off-by: Ahmed Ismail <Ahmed.Ismail@arm.com>
The function vApplicationFPUSafeIRQHandler gets the value of ICCIAR as
parameter, but a constant containing the address of ICCIAR was also
defined. Fix the name of the constant to align it with what it actually
holds.
* Pass core ID to CRITICAL_NESTING_COUNT macros
* Match existing data type for xCoreID
* Get core ID when interrupts are disabled
* Implement get core ID with interrupt disabled
* Get core ID inline within vTaskSuspendAll() to resolve compiler warning
* Fix formatting check
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
---------
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
Co-authored-by: Ching-Hsin,Lee <chinglee@amazon.com>
Co-authored-by: Gaurav Aggarwal <aggarg@amazon.com>
Co-authored-by: Gaurav-Aggarwal-AWS <33462878+aggarg@users.noreply.github.com>
There is a regression issue introduced in
cfc268814a.
This PR was intended to update MIT licensed header from v9 to v10.
But it accidentally changed "portYIELD_FROM_ISR( x )" in MSVC-MingW/portmacro.h.
It caused "portYIELD_FROM_ISR( x )" does not return correct value to "prvProcessSimulatedInterrupts".
Add public dependency to hardware_sync because portable\ThirdParty\GCC\RP2040\include\portmacro.h exposes include "hardware/sync.h" to FreeRTOS.h which is PUBLIC.
Add private dependency to pico_multicore because of usage prvFIFOInterruptHandler() and sio_hw in portable/ThirdParty/GCC/RP2040/port.c.
The PACBTI is an optional hardware security feature,
the current implementation assumes that every SoC that
has Armv8.1-M architecture extension, has the PACBTI
hardware feature, which does not have to be the case.
Hence, the `portHAS_PACBTI_FEATURE` is removed
and the implementation is modified to rely on
`configENABLE_PAC` and `configENABLE_BTI` macros
that can either be set using CMake or FreeRTOSConfig.h
header file.
Enabling PAC and/or BTI on a port variant that
doesn't have the PACBTI hardware feature would be
caught by a `configASSERT` statement.
Signed-off-by: Ahmed Ismail <Ahmed.Ismail@arm.com>
Co-authored-by: Tony Josi <tonyjosi@amazon.com>
