Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix xTaskNotifyWait & ulTaskNotifyTake determinism. #625

Closed
wants to merge 2 commits into from
Closed
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
80 changes: 46 additions & 34 deletions tasks.c
Original file line number Diff line number Diff line change
Expand Up @@ -3137,8 +3137,8 @@ void vTaskPlaceOnEventList( List_t * const pxEventList,
{
configASSERT( pxEventList );

/* THIS FUNCTION MUST BE CALLED WITH EITHER INTERRUPTS DISABLED OR THE
* SCHEDULER SUSPENDED AND THE QUEUE BEING ACCESSED LOCKED. */
/* THIS FUNCTION MUST BE CALLED WITH THE SCHEDULER SUSPENDED AND THE QUEUE
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why was this comment changed? The original "Interrupts disabled" should probably be "inside a critical section", but I think it's still valid.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I changed it because this function calls vListInsert(), which shouldn't be called while interrupts are disabled as far as I am aware. I noted this comment in my original issue. But I can change it back if you think this is not correct.

* BEING ACCESSED LOCKED. */

/* Place the event list item of the TCB in the appropriate event list.
* This is placed in the list in priority order so the highest priority task
Expand Down Expand Up @@ -4744,22 +4744,25 @@ TickType_t uxTaskResetEventItemValue( void )
configASSERT( uxIndexToWait < configTASK_NOTIFICATION_ARRAY_ENTRIES );

taskENTER_CRITICAL();

/* Only block if the notification count is not already non-zero. */
if( pxCurrentTCB->ulNotifiedValue[ uxIndexToWait ] == 0UL )
{
/* Only block if the notification count is not already non-zero. */
if( pxCurrentTCB->ulNotifiedValue[ uxIndexToWait ] == 0UL )
/* Mark this task as waiting for a notification. */
pxCurrentTCB->ucNotifyState[ uxIndexToWait ] = taskWAITING_NOTIFICATION;

if( xTicksToWait > ( TickType_t ) 0 )
{
/* Mark this task as waiting for a notification. */
pxCurrentTCB->ucNotifyState[ uxIndexToWait ] = taskWAITING_NOTIFICATION;
traceTASK_NOTIFY_TAKE_BLOCK( uxIndexToWait );

if( xTicksToWait > ( TickType_t ) 0 )
{
prvAddCurrentTaskToDelayedList( xTicksToWait, pdTRUE );
traceTASK_NOTIFY_TAKE_BLOCK( uxIndexToWait );
/* Suspend the scheduler before enabling interrupts. */
vTaskSuspendAll();
taskEXIT_CRITICAL();
Comment on lines +4758 to +4760
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Strictly speaking, the API doesn't allow calling vTaskSuspendAll() from inside a critical section. It probably works, but it would be the only place it happens in the FreeRTOS codebase.

The intent behind the larger construct being proposed here seems to be to avoid suspending/resuming the scheduler if possible, which in turn keeps ulTaskGenericNotifyTake() as lightweight as possible. Maybe a different construct could be used instead. For example, the idle task uses a different construct to avoid suspending/resuming the scheduler for tickless idle. A similar approach might work here -- specifically, checking the notification value while "unprotected" to determine whether to suspend the scheduler, and then verifying the notification value after suspending the scheduler if suspending is warranted.


prvAddCurrentTaskToDelayedList( xTicksToWait, pdTRUE );

/* All ports are written to allow a yield in a critical
* section (some will yield immediately, others wait until the
* critical section exits) - but it is not something that
* application code should ever do. */
if( xTaskResumeAll() == pdFALSE )
{
portYIELD_WITHIN_API();
}
else
Expand All @@ -4769,10 +4772,13 @@ TickType_t uxTaskResetEventItemValue( void )
}
else
{
mtCOVERAGE_TEST_MARKER();
taskEXIT_CRITICAL();
}
}
taskEXIT_CRITICAL();
else
{
taskEXIT_CRITICAL();
}

taskENTER_CRITICAL();
{
Expand Down Expand Up @@ -4818,27 +4824,30 @@ TickType_t uxTaskResetEventItemValue( void )
configASSERT( uxIndexToWait < configTASK_NOTIFICATION_ARRAY_ENTRIES );

taskENTER_CRITICAL();

/* Only block if a notification is not already pending. */
if( pxCurrentTCB->ucNotifyState[ uxIndexToWait ] != taskNOTIFICATION_RECEIVED )
{
/* Only block if a notification is not already pending. */
if( pxCurrentTCB->ucNotifyState[ uxIndexToWait ] != taskNOTIFICATION_RECEIVED )
/* Clear bits in the task's notification value as bits may get
* set by the notifying task or interrupt. This can be used to
* clear the value to zero. */
pxCurrentTCB->ulNotifiedValue[ uxIndexToWait ] &= ~ulBitsToClearOnEntry;

/* Mark this task as waiting for a notification. */
pxCurrentTCB->ucNotifyState[ uxIndexToWait ] = taskWAITING_NOTIFICATION;

if( xTicksToWait > ( TickType_t ) 0 )
{
/* Clear bits in the task's notification value as bits may get
* set by the notifying task or interrupt. This can be used to
* clear the value to zero. */
pxCurrentTCB->ulNotifiedValue[ uxIndexToWait ] &= ~ulBitsToClearOnEntry;
traceTASK_NOTIFY_WAIT_BLOCK( uxIndexToWait );

/* Mark this task as waiting for a notification. */
pxCurrentTCB->ucNotifyState[ uxIndexToWait ] = taskWAITING_NOTIFICATION;
/* Suspend the scheduler before enabling interrupts. */
vTaskSuspendAll();
taskEXIT_CRITICAL();

if( xTicksToWait > ( TickType_t ) 0 )
{
prvAddCurrentTaskToDelayedList( xTicksToWait, pdTRUE );
traceTASK_NOTIFY_WAIT_BLOCK( uxIndexToWait );
prvAddCurrentTaskToDelayedList( xTicksToWait, pdTRUE );

/* All ports are written to allow a yield in a critical
* section (some will yield immediately, others wait until the
* critical section exits) - but it is not something that
* application code should ever do. */
if( xTaskResumeAll() == pdFALSE )
{
portYIELD_WITHIN_API();
}
else
Expand All @@ -4848,10 +4857,13 @@ TickType_t uxTaskResetEventItemValue( void )
}
else
{
mtCOVERAGE_TEST_MARKER();
taskEXIT_CRITICAL();
}
}
taskEXIT_CRITICAL();
else
{
taskEXIT_CRITICAL();
}

taskENTER_CRITICAL();
{
Expand Down