chinglee-iot [Tue, 24 Jan 2023 08:25:36 +0000 (16:25 +0800)]
Remove prvSelectHighestPriorityTask call in vTaskSuspend (#610)
* Remove prvSelectHighestPriorityTask call in vTaskSuspend
* Every core starts with an idle task in SMP implementation and
taskTASK_IS_RUNNING only return ture when the task is idle task before
scheduler started. So prvSelectHighestPriorityTask won't be called in
vTaskSuspend before scheduler started.
* Update prvSelectHighestPriorityTask to ensure that this function is
called only when scheduler started.
Sudeep Mohanty [Wed, 22 Jun 2022 04:33:44 +0000 (06:33 +0200)]
Update prvYieldCore() compile warning for single core targets (#505)
When configNUM_CORES is 1, prvYieldCore() generates an array subscript
outofbound error (-Warray-bounds) when compiled with GCC with space
optimization enabled (-Os).
This commit updates the code flow in prvYieldCore() to compile out
the part where yield is needed on the other core which is unnecessary
for single-core targets.
Darian [Tue, 31 May 2022 23:11:02 +0000 (07:11 +0800)]
Add support for newlib dynamic reentrancy (#496)
Previously, newlib's _impure_ptr was updated on every context switch
to point to the current task's _reent structure.
However, this behavior is no longer valid on multi-core systems due
to the fact that multiple cores can switch contexts at the same time,
thus leading to the corruption of the _impure_ptr.
However, Newlib can be compiled with __DYNAMIC_REENT__ enabled which
will cause newlib functions to call __getreent() instead in order to
obtain the required reent struct.
This commit adds dynamic reentrancy support to FreeRTOS:
- Added a configNEWLIB_REENTRANT_IS_DYNAMIC to enable dynamic reentrancy support
- _impure_ptr is no longer updated with reentrancy is dynamic
- Port must provide their own __getreent() that returns the current task's reent struct
Timo Sandmann [Tue, 28 Dec 2021 20:30:03 +0000 (21:30 +0100)]
smp branch: bugfix for race condition on RP2040 (#431)
* Bugfix for race condition on RP2040 in vPortEnableInterrupts()
RP2040 SMP port: Since spin_unlock() re-enables interrupts, pxYieldSpinLock has to be updated first to avoid a possible race condition.
* Bugfix for invalid sanity checks on RP2040
RP2040 SMP port: Testing pxYieldSpinLock for NULL does not work reliable in these places, because another/new lock might already be set when configASSERT() is executed.
Michael Bruno [Mon, 28 Jun 2021 15:44:24 +0000 (11:44 -0400)]
Fix XCore DP/CP register corruption bug (#352)
The kernel code that runs within RTOS ISRs requires that DP and CP not
be changed from their initial values. If a task changes them and is
interrupted while they are changed, then they must be restored for the
ISR. This commit implements this.
This relies on a corresponding change to the xcore lib_rtos_support.
Joseph Julicher [Tue, 16 Mar 2021 20:57:22 +0000 (13:57 -0700)]
SMP Documentation (#280)
* Add XMOS XCore ports
Some minor modifications are also made to the kernel to support the
XCore compiler's automatic stack size calculation.
* Update kernel to support SMP
The XMOS XCore ports are also updated to support SMP.
* Fix compiler warnings in xcore ports
The port set and clear interrupt mask from ISR macros were removed from
the ports so that the default macros found in FreeRTOS.h are used
instead. The default macros do not result in warnings when they are
used.
* Remove inline function from timers.h
Inline function converted to macro. This should now build when
optimizations are off and inlining is disabled.
* Fix compiler warnings in xcore ports and tasks.c
* fixed documentation for ulTaskNotifyTake() and ulTaskNotifyTakeIndexed()
Joseph Julicher [Tue, 16 Mar 2021 00:29:08 +0000 (17:29 -0700)]
Feature: SMP (#278)
* Add XMOS XCore ports
Some minor modifications are also made to the kernel to support the
XCore compiler's automatic stack size calculation.
* Update kernel to support SMP
The XMOS XCore ports are also updated to support SMP.
* Fix compiler warnings in xcore ports
The port set and clear interrupt mask from ISR macros were removed from
the ports so that the default macros found in FreeRTOS.h are used
instead. The default macros do not result in warnings when they are
used.
* Remove inline function from timers.h
Inline function converted to macro. This should now build when
optimizations are off and inlining is disabled.
* Fix compiler warnings in xcore ports and tasks.c
* fixed documentation for ulTaskNotifyTake() and ulTaskNotifyTakeIndexed()
Gaurav-Aggarwal-AWS [Fri, 12 Mar 2021 20:11:19 +0000 (12:11 -0800)]
Update comment in list.c (#276)
The tick interrupt priority must be configured to ensure that it cannot
interrupt a critical section. This change updates the comment to help
the application writer while debugging.
Paul Bartell [Sat, 6 Mar 2021 02:46:49 +0000 (18:46 -0800)]
Add some assertions and coverage exceptions to queue.c (#273)
* Add an LCOV_BRANCH exception for the check that sizeof( StaticQueue_t ) == sizeof( Queue_t )
* Add LCOV_BRANCH coverage exception for a configASSERT on pxQueueSetContainer with a condition that is unreachable.
* Add configASSERTs to alert when invalid parameters are passed into Queue Registry related functions.
* Assert that the semaphore handle passed into xQueueGetMutexHolder is not NULL.
* Correct some typos in queue.c
* Update lexicon.txt
Gaurav-Aggarwal-AWS [Wed, 24 Feb 2021 02:36:27 +0000 (18:36 -0800)]
Support allocating stack from separate heap (#267)
The change adds support for allocating task stacks from separate heap.
When configSTACK_ALLOCATION_FROM_SEPARATE_HEAP is defined as 1 in
FreeRTOSConfig.h, task stacks are allocated and freed using
pvPortMallocStack and vPortFreeStack functions. This allows the
application writer to provide a separate allocator for task stacks.
When configSTACK_ALLOCATION_FROM_SEPARATE_HEAP is defined as 0, task
stacks are allocated and freed using FreeRTOS heap functions
pvPortMalloc and vPortFree.
For backward compatibility, configSTACK_ALLOCATION_FROM_SEPARATE_HEAP
defaults to 0.
Gaurav-Aggarwal-AWS [Mon, 7 Dec 2020 17:53:22 +0000 (09:53 -0800)]
Prevent unprivileged task from altering MPU configuration (#227)
This change removes the FreeRTOS System Calls (aka MPU wrappers) for the
following kernel APIs:
- xTaskCreateRestricted
- xTaskCreateRestrictedStatic
- vTaskAllocateMPURegions
A system call allows an unprivileged task to execute a kernel API which
is otherwise accessible to privileged software only. The above 3 APIs
can create a new task with a different MPU configuration or alter the
MPU configuration of an existing task. This an be (mis)used by an
unprivileged task to grant itself access to a region which it does not
have access to.
Removing the system calls for these APIs ensures that an unprivileged
task cannot execute this APIs. If an unprivileged task attempts to
execute any of these API, it will result in a Memory Fault.
uxCriticalNesting is initialized to a large value at the start and set
to zero when the scheduler is started (xPortStartScheduler). As a
result, before the scheduler is started, a pair of enter/exit critical
section will leave the interrupts disabled because uxCriticalNesting
will not reach zero in the vPortExitCritical function. This is done to
ensure that the interrupts remain disabled from the time first FreeRTOS
API is called to the time when the scheduler is started. The scheduler
starting code is expected to enure that interrupts are enabled before
the first task starts executing.
Cortex-M33 ports were not enabling interrupts before starting the first
task and as a result, the first task was started with interrupts
disabled. This PR fixes the issue by ensuring that interrupts are
enabled before the first task is started.
magicse7en [Mon, 26 Oct 2020 18:47:21 +0000 (02:47 +0800)]
Xtensa: fix the coproc_area incorrect issue (#117)
* Xtensa: fix the coproc_area incorrect issue
foss-xtensa/amazon-freertos#2 mentioned a issue:
1.
In function pxPortInitialiseStack(StackType_t *pxTopOfStack....)
p = (uint32_t *)(((uint32_t) pxTopOfStack - XT_CP_SIZE) & ~0xf);
In function prvInitialiseNewTask (file: task.c)
pxTopOfStack = (pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)
So the co-processor area is at
p = (uint32_t *)(((uint32_t)((pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)) - XT_CP_SIZE) & ~0xf);
2.
In function vPortStoreTaskMPUSettings( .... , StackType_t pxBottomOfStack ...)
xMPUSettings->coproc_area = (StackType_t)((((uint32_t)(pxBottomOfStack + usStackDepth - 1)) - XT_CP_SIZE) & ~0xf);
The p is coproc_area that should be equal to xMPUSettings->coproc_area.
For example, assume pxStack is 0xa0000000, ulStackDepth is 0x2000,
portBYTE_ALIGNMENT_MASK is 0x7f, XT_CP_SIZE is 0x100.
The p = (uint32_t)(((uint32_t)((pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)) - XT_CP_SIZE) & ~0xf)
= 0xa0001e80
The xMPUSettings->coproc_area = (StackType_t)((((uint32_t)(pxStack+ usStackDepth - 1)) - XT_CP_SIZE ) & ~0xf)
= 0xa0001ef0
Obviously, the p is not equal to the xMPUSettings->coproc_area, which will cause context switching error.
begriffs.com / projects / freertos / log