Disassembler seems to swallow the corrupted part of code without error against corrupted class file

[HeartSaVioR]

Migrated from janino-compiler/janino#113 - the original issue was fixed, but while debugging, disassembler doesn't give any information against corrupted class file hence had to figure out with the field code (byte array) against debugger.

Corrupted class file was generated from Janino (the reason was janino-compiler/janino#113) - I've set breakpoint to UnitCompiler.addClassFile and force debugger to write cf.byteArray() to the file, as well as run Disassembler.disassembleToStdout(cf.toByteArray()).

Archive.zip

• error-codegen.log: The code triggered the bug Padding for TABLESWITCH can be applied multiple times due to grow on relocatables janino-compiler/janino#113

• classfile.class: The content cf.byteArray() is stored.

• classfile.class.decompiled-javac: The output from javap -c -p -v classfile.class

It threw error and stopped printing out disassembled code where the bug has been spotted. (tableswitch is in offset 578 in expand_doConsume - got the information while debugging with code field)

Error: A serious internal error has occurred: java.lang.NegativeArraySizeException
Please file a bug report, and include the following information:
java.lang.NegativeArraySizeException
	at com.sun.tools.classfile.Instruction.accept(Instruction.java:288)
	at com.sun.tools.javap.CodeWriter.writeInstr(CodeWriter.java:126)
	at com.sun.tools.javap.CodeWriter.writeInstrs(CodeWriter.java:108)
	at com.sun.tools.javap.CodeWriter.write(CodeWriter.java:75)
	at com.sun.tools.javap.AttributeWriter.visitCode(AttributeWriter.java:220)
	at com.sun.tools.javap.AttributeWriter.visitCode(AttributeWriter.java:75)
	at com.sun.tools.classfile.Code_attribute.accept(Code_attribute.java:102)
	at com.sun.tools.javap.AttributeWriter.write(AttributeWriter.java:101)
	at com.sun.tools.javap.ClassWriter.writeMethod(ClassWriter.java:508)
	at com.sun.tools.javap.ClassWriter.writeMethods(ClassWriter.java:412)
	at com.sun.tools.javap.ClassWriter.write(ClassWriter.java:217)
	at com.sun.tools.javap.JavapTask.write(JavapTask.java:784)
	at com.sun.tools.javap.JavapTask.writeClass(JavapTask.java:606)
	at com.sun.tools.javap.JavapTask.run(JavapTask.java:560)
	at com.sun.tools.javap.JavapTask.run(JavapTask.java:423)
	at com.sun.tools.javap.Main.main(Main.java:46)

• classfile.class.decompiled-jdisasm-1.x: The output from Disassembler.disassembleToStdout(cf.toByteArray()) (I've done with two versions: 1.4 and 1.5)

• classfile.class.decompiled-jdisasm-1.5-cli: The output from de.unkrig.jdisasm.Disassembler -verbose

All outputs from jdisasm silently threw out the part (entire code in the method - expand_doConsume) where corrupted operation is placed.

As jdisasm is the recommended way to report the bug on Janino, it would be nice to catch the corrupted code and throw out error (or tolerate the error but log the error message at least).