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.
The function to change a thread priority was too complex. Simplify it
with a new scheduler operation. This increases the average case
performance due to the simplified logic. The interrupt disabled
critical section is a bit prolonged since now the extract, update and
enqueue steps are executed atomically. This should however not impact
the worst-case interrupt latency since at least for the Deterministic
Priority Scheduler this sequence can be carried out with a wee bit of
instructions and no loops.
Add _Scheduler_Change_priority() to replace the sequence of
- _Thread_Set_transient(),
- _Scheduler_Extract(),
- _Scheduler_Enqueue(), and
- _Scheduler_Enqueue_first().
Delete STATES_TRANSIENT, _States_Is_transient() and
_Thread_Set_transient() since this state is now superfluous.
With this change it is possible to get rid of the
SCHEDULER_SMP_NODE_IN_THE_AIR state. This considerably simplifies the
implementation of the new SMP locking protocols.
The current implementation of task migration in RTEMS has some
implications with respect to the interrupt latency. It is crucial to
preserve the system invariant that a task can execute on at most one
processor in the system at a time. This is accomplished with a boolean
indicator in the task context. The processor architecture specific
low-level task context switch code will mark that a task context is no
longer executing and waits that the heir context stopped execution
before it restores the heir context and resumes execution of the heir
task. So there is one point in time in which a processor is without a
task. This is essential to avoid cyclic dependencies in case multiple
tasks migrate at once. Otherwise some supervising entity is necessary to
prevent life-locks. Such a global supervisor would lead to scalability
problems so this approach is not used. Currently the thread dispatch is
performed with interrupts disabled. So in case the heir task is
currently executing on another processor then this prolongs the time of
disabled interrupts since one processor has to wait for another
processor to make progress.
It is difficult to avoid this issue with the interrupt latency since
interrupts normally store the context of the interrupted task on its
stack. In case a task is marked as not executing we must not use its
task stack to store such an interrupt context. We cannot use the heir
stack before it stopped execution on another processor. So if we enable
interrupts during this transition we have to provide an alternative task
independent stack for this time frame. This issue needs further
investigation.
Use register g6 for the per-CPU control of the current processor. The
register g6 is reserved for the operating system by the SPARC ABI. On
Linux register g6 is used for a similar purpose with the same method
since 1996.
The register g6 must be initialized during system startup and then must
remain unchanged.
Since the per-CPU control is used in all critical sections of the
operating system, this is a performance optimization for the operating
system core procedures. An additional benefit is that the low-level
context switch and interrupt processing code is now identical on non-SMP
and SMP configurations.
The registers g2 through g4 are reserved for applications. GCC uses
them as volatile registers by default. So they are treated like
volatile registers in RTEMS as well.
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.
Do not allocate the scheduler control structures from the workspace.
This is a preparation step for configuration of clustered/partitioned
schedulers on SMP.
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.
Use posix keys for current shell environment instead of task variables. With
this patch the shell needs one posix-key and one posix-key-value-pair
configured.
Update documentation for the shell.
Adapt samples/fileio:
- Add necessary objects.
- Add login function and custom device name for better testing of the shell.
