8.14.1. Capturing trace after an event and stopping

Two things must happen before trace capture can be stopped:

Figure 8.21 shows a possible timeline of events where an event of interest, referred to as a trigger event, causes some trace that must be captured and thereafter the trace capture device can be stopped.

When one trace source is used, there is no requirement to flush the system, instead the length of the trigger counter delay can be increased to enable more trace to be generated, thereby pushing out historical information.

Traditionally only the initial trigger event is sent to the TPA at time t1, indicated using TRACECTL and a special encoding on TRACEDATA. This can still be done but if trace is stopped at this point, there might still be related trace stalled within the ATB system. Trigger signals are now abstracted from ATB through the CTI/CTM infrastructure. To allow all trace information to have been output that could have related to an internally generated trigger event, the system must be flushed after which trace capture can be safely stopped.

Figure 8.21. Capturing trace after an event and stopping


In Figure 8.21 the action to cause trace capture to be stopped at time t3 could be:

Copyright © 2004-2009 ARM. All rights reserved.ARM DDI 0314H
Non-Confidential