With -XX:+PrintInterpreter the instructions of the generated interpreter are printed at startup.

But it is possible that after startup other traces are flooded with repeatedly decoded interpreter sections.

For instance -Xlog:continuations=trace suffers from this.

Reproduce:

- Download the attached Simple.java

- run the command
  java -XX:+PrintInterpreter -Xlog:continuations=trace Simple.java > out.log
  
- Search for "Interpreted frame" in out.log

- You will see that at each occurence the "return entry points" are decoded again and again

This is a stacktrace where the repeated printing happens:

InterpreterCodelet::print_on(outputStream *)
frame::print_value_on(outputStream *)
FreezeBase::after_freeze_java_frame(const frame &, bool)

InterpreterCodelet::print_on(outputStream *) decodes the instructions repeadedly because the decoding depends directly on PrintInterpreter.

The attached partial_out.log is an excerpt from the continuations trace. The decoded "return entry points" codelet is a significant part of the trace. It occurs repeatedly and is redundant because the interpreter was printed at startup.