Use the PTHREAD mutexes and condition variables if available. This
helps on SMP configurations to avoid the home grown condition variables
via disabled preemption.
Add basic support for the Multiprocessor Resource Sharing Protocol
(MrsP).
The Multiprocessor Resource Sharing Protocol (MrsP) is defined in A.
Burns and A.J. Wellings, A Schedulability Compatible Multiprocessor
Resource Sharing Protocol - MrsP, Proceedings of the 25th Euromicro
Conference on Real-Time Systems (ECRTS 2013), July 2013. It is a
generalization of the Priority Ceiling Protocol to SMP systems. Each
MrsP semaphore uses a ceiling priority per scheduler instance. These
ceiling priorities can be specified with rtems_semaphore_set_priority().
A task obtaining or owning a MrsP semaphore will execute with the
ceiling priority for its scheduler instance as specified by the MrsP
semaphore object. Tasks waiting to get ownership of a MrsP semaphore
will not relinquish the processor voluntarily. In case the owner of a
MrsP semaphore gets preempted it can ask all tasks waiting for this
semaphore to help out and temporarily borrow the right to execute on one
of their assigned processors.
The help out feature is not implemented with this patch.
Rename scheduler per-thread information into scheduler nodes using
Scheduler_Node as the base type. Use inheritance for specialized
schedulers.
Move the scheduler specific states from the thread control block into
the scheduler node structure.
Validate the SMP scheduler node state transitions in case RTEMS_DEBUG is
defined.
Clustered/partitioned scheduling helps to control the worst-case
latencies in the system. The goal is to reduce the amount of shared
state in the system and thus prevention of lock contention. Modern
multi-processor systems tend to have several layers of data and
instruction caches. With clustered/partitioned scheduling it is
possible to honour the cache topology of a system and thus avoid
expensive cache synchronization traffic.
We have clustered scheduling in case the set of processors of a system
is partitioned into non-empty pairwise-disjoint subsets. These subsets
are called clusters. Clusters with a cardinality of one are partitions.
Each cluster is owned by exactly one scheduler instance.
The thread control block contains fields that point to application
configuration dependent memory areas, like the scheduler information,
the API control blocks, the user extension context table, the RTEMS
notepads and the Newlib re-entrancy support. Account for these areas in
the configuration and avoid extra workspace allocations for these areas.
This helps also to avoid heap fragementation and reduces the per thread
memory due to a reduced heap allocation overhead.
Do not allocate the scheduler control structures from the workspace.
This is a preparation step for configuration of clustered/partitioned
schedulers on SMP.
Rename rtems_smp_get_current_processor() in
rtems_get_current_processor(). Make rtems_get_current_processor() a
function in uni-processor configurations to enable ABI compatibility
with SMP configurations.
Rename rtems_smp_get_processor_count() in rtems_get_processor_count().
Make rtems_get_processor_count() a function in uni-processor
configurations to enable ABI compatibility with SMP configurations.
This simplifies the RTEMS initialization and helps to avoid a memory
overhead. The workspace demands of the IO manager were not included in
the <rtems/confdefs.h> workspace size estimate. This is also fixed as a
side-effect.
Update documentation and move "Specifying Application Defined Device
Driver Table" to the section end. This sub-section is not that
important for the user. Mentioning this at the beginning may lead to
confusion.
Per task variables are inherently unsafe in SMP systems. This
patch disables them from the build and adds warnings in the
appropriate documentation and configuration sections.
Delete global variables _Priority_Major_bit_map and _Priority_Bit_map.
This makes it possible to use multiple priority scheduler instances for
example with clustered/partitioned scheduling on SMP.
Issue a fatal error in case a thread is deleted which still owns
resources (e.g. a binary semaphore with priority inheritance or ceiling
protocol). The resource count must be checked quite late since RTEMS
task variable destructors, POSIX key destructors, POSIX cleanup handler,
the Newlib thread termination extension or other thread termination
extensions may release resources. In this context it would be quite
difficult to return an error status to the caller.
An alternative would be to place threads with a non-zero resource count
not on the zombie chain. Thus we have a resource leak instead of a
fatal error. The terminator thread can see this error if we return an
RTEMS_RESOURCE_IN_USE status for the rtems_task_delete() for example.
Add per-CPU profiling stats API. Implement the thread dispatch disable
level profiling. The interrupt profiling must be implemented in CPU
port specific parts (mostly assembler code). Add a support function
_Profiling_Outer_most_interrupt_entry_and_exit() for this purpose.
Add a local context structure to the SMP lock API for acquire and
release pairs. This context can be used to store the ISR level and
profiling information. It may be later used to enable more
sophisticated lock algorithms, e.g. MCS locks.
There is only one lock that cannot be used with a local context. This
is the per-CPU lock since here we would have to transfer the local
context through a context switch which is very complicated.
This partially reverts commit 1215fd4d94.
In order to support profiling of SMP locks and provide a future
compatible SMP locks API it is necessary to add an SMP lock destroy
function. Since the commit above adds an SMP lock to each chain control
we would have to add a rtems_chain_destroy() function as well. This
complicates the chain usage dramatically. Thus revert the patch above.
A global SMP lock for all chains is used to implement the protected
chain operations.
Advantages:
* The SAPI chain API is now identical on SMP and non-SMP
configurations.
* The size of the chain control is reduced and is then equal to the
Score chains.
* The protected chain operations work correctly on SMP.
Disadvantage:
* Applications using many different chains and the protected operations
may notice lock contention.
The chain control size drop is a huge benefit (SAPI chain controls are
66% larger than the Score chain controls). The only disadvantage is not
really a problem since these applications can use specific interrupt
locks and unprotected chain operations to avoid this issue.
Formerly POSIX keys were only enabled when POSIX threads
were enabled. Because they are a truly safe alternative
to per-task variables in an SMP system, they are being
enabled in all configurations.
SMP must be initialized in order to know the current set of
cores available. Without this, you cannot initialize the
default cpu_set_t associated with Classic API tasks and
POSIX threads.
Rename _SMP_Request_other_cores_to_perform_first_context_switch() into
_SMP_Request_start_multitasking() since this requests now a multitasking
start on all configured and available processors. The name corresponds
_Thread_Start_multitasking() and
_SMP_Start_multitasking_on_secondary_processor() actions issued in
response to this request. Move in source file to right place.
Rename PER_CPU_STATE_READY_TO_BEGIN_MULTITASKING into
PER_CPU_STATE_READY_TO_START_MULTITASKING.
Rename PER_CPU_STATE_BEGIN_MULTITASKING into
PER_CPU_STATE_REQUEST_START_MULTITASKING.
Rename _SMP_Request_other_cores_to_shutdown() into
_SMP_Request_shutdown().
Add a per-CPU state lock to protect all changes. This was necessary to
offer a controlled shutdown of the system (atomic read/writes alone are
not sufficient for this kind of synchronization).
Add documentation for Per_CPU_State.
Delete debug output.
New tests smptests/smpfatal01 and smptests/smpfatal02.
