This report came from Werner Dietl:

I have a question about the type_index field in TypeAnnotationPosition.java:

http://hg.openjdk.java.net/jdk8/jdk8/langtools/file/1ff9d5118aae/src/share/classes/com/sun/tools/javac/code/TypeAnnotationPosition.java#l147
http://hg.openjdk.java.net/jdk9/jdk9/langtools/file/3665ebc22a42/src/jdk.compiler/share/classes/com/sun/tools/javac/code/TypeAnnotationPosition.java#l147

For class extends clauses this field will be -1 and for interface
implements clauses the index will correspond to the index of the
implemented interface.

The -1 is written as two bytes, 0xFFFF, at:

http://hg.openjdk.java.net/jdk8/jdk8/langtools/file/1ff9d5118aae/src/share/classes/com/sun/tools/javac/jvm/ClassWriter.java#l948
http://hg.openjdk.java.net/jdk9/jdk9/langtools/file/3665ebc22a42/src/jdk.compiler/share/classes/com/sun/tools/javac/jvm/ClassWriter.java#l906

However, when the 0xFFFF bytes are read back in, at:

http://hg.openjdk.java.net/jdk8/jdk8/langtools/file/1ff9d5118aae/src/share/classes/com/sun/tools/javac/jvm/ClassReader.java#l1566
http://hg.openjdk.java.net/jdk9/jdk9/langtools/file/3665ebc22a42/src/jdk.compiler/share/classes/com/sun/tools/javac/jvm/ClassReader.java#l1683

the type index is set to 65535.

On the other hand, at:

http://hg.openjdk.java.net/jdk8/jdk8/langtools/file/1ff9d5118aae/src/share/classes/com/sun/tools/classfile/TypeAnnotation.java#l134
http://hg.openjdk.java.net/jdk9/jdk9/langtools/file/3665ebc22a42/src/jdk.jdeps/share/classes/com/sun/tools/classfile/TypeAnnotation.java#l134

the value 0xFFFF is special-cased and replaced by -1.

As javap is using the correct version from
classfile/TypeAnnotation.java I don't see a simple way to illustrate
the issue.
If you write an annotation processor that sees a class extends
annotation, you would observe 65535 instead of -1.

I think the mismatch between jvm/ClassReader and
classfile/TypeAnnotation must be fixed to ensure a consistent
interpretation.