Use RTEMS_SYSINIT_ORDER_LAST_BUT_5 instead of RTEMS_SYSINIT_ORDER_LAST
to allow applications and support functions to place system
initialization handlers behind the standard handlers.
Update #3838.
Introduce new library librtemsrecordwrap.a which contains wrappers for
operating system functions which produce entry/exit events.
The wrappers can be selected during link time via the GNU ld --wrap
option.
Update #3665.
Reduce the system dependencies to allow tracing of very low level
functions, for example the interrupt disable/enable.
Introduce general purpose RTEMS_RECORD_CALLER and RTEMS_RECORD_LINE
events.
Update #3665.
The .rtemsrwset section is used for the per-CPU data. This section has
loadable content. Place the ring buffers in the BSS section to avoid
large executable image sizes.
Not using the per-CPU data makes it possible to initialize the record
support earlier.
Update #3665.
In case of a ring buffer overflow, the rtems_record_drain() will push
the complete ring buffer content to the client. While the items are
processed by the client, new items may overwrite some items being
processed. The overwritten items can be detected in the following
iteration once the next tail/head information is pushed to the client.
Add rtems_scheduler_get_processor_maximum() as a replacement for
rtems_get_processor_count(). The rtems_get_processor_count() is a bit
orphaned. Adopt it by the Scheduler Manager. The count is also
misleading, since the processor set may have gaps and the actual count
of online processors may be less than the value returned by
rtems_get_processor_count().
Update #3732.
Add low level event recording infrastructure for system and user
defined events. The infrastructure is able to record high frequency
events such as
* SMP lock acquire/release,
* interrupt entry/exit,
* thread switches,
* UMA zone allocate/free, and
* Ethernet packet input/output, etc.
It allows post-mortem analysis in fatal error handlers, e.g. the last
events are in the record buffer, the newest event overwrites the oldest
event. It is possible to detect record buffer overflows for consumers
that expect a continuous stream of events, e.g. to display the system
state in real-time.
The implementation supports high-end SMP machines (more than 1GHz
processor frequency, more than four processors).
Add a new API instead. The implementation uses per-processor data
structures and no atomic read-modify-write operations. It is uses
per-processor ring buffers to record the events.
The CPU counter is used to get the time of events. It is combined with
periodic uptime events to synchronize it with CLOCK_REALTIME.
The existing capture engine tries to solve this problem also, but its
performance is not good enough for high-end production systems. The
main issues are the variable-size buffers and the use of SMP locks for
synchronization. To fix this, the API would change significantly.
Update #3665.
