It looks like the function

bool OopMapCacheEntry::verify_mask(CellTypeState* vars, CellTypeState* stack, int max_locals, int stack_top) {
  . . .
  Log(interpreter, oopmap) logv;
  LogStream st(logv.trace());

  st.print("Locals (%d): ", max_locals);
  for(int i = 0; i < max_locals; i++) {
    . . .
    st.print("%d", v1 ? 1 : 0);
  }
  st.cr();

  st.print("Stack (%d): ", stack_top);
  for(int j = 0; j < stack_top; j++) {
    . . .
    st.print("%d", v1 ? 1 : 0);
  }
  st.cr();
  return true;
}

was written in wrong belief that if logging is disabled for specific target than associated LogStream::print() does noting. That is a mistake. At first the control flow goes to outputStream::print() that calls sprintf(), then it goes to LogTagSet::vwrite() that calls vsnprintf(), and only after that in LogTagSet::log() it finds that the list of LogOutput's is empty.

I developed a kinda of logging improvement when found that resulting log is full of [interpreter,oopmap] messages in spite of the fact that I configured only GC logging (see attached log file). I'd suggest to use regular log_trace() macro that checks the necessity before to generate a log message