This mail to jigsaw-dev suggests there are issues in the jimage and image reader code when using compression:

http://mail.openjdk.java.net/pipermail/jigsaw-dev/2018-August/013905.html

These issues should be investigated, there may be some defensive-in-depth changes needed to cope with a buggy writer. Also there may be some performance work needed when using compression (the JDK run-time image doesn't use compress so it's really only an issue for someone creating a run-time image with `jlink --compress=<level>`.