This is the mail archive of the java@gcc.gnu.org mailing list for the Java project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: need help debugging


>>>>> "Michael" == Michael Koch <konqueror@gmx.de> writes:

>> You're going to have to use `jcf-dump -c' to get the JVM source code
>> and then see if PC 305 in that file really is bad bytecode.

Michael> 305: aload 4

Michael> I dont know if "aload 4" is just illegal or noninterpreted by
Michael> gij or none of this. Can somebody enlighten me on this ?

Unfortunately seeing if this is a bug in the bytecode or a bug in the
libgcj verifier isn't always very easy.  Usually you have to either
debug the verifier or read the bytecode pretty closely, following the
logic of it.  This is much easier if the program is shorter.

Getting this error from an `aload' means that libgcj thinks that local
variable 4 doesn't hold an object reference at this point in time.

I might be able to debug it if:

* You sent all the bytecode for the method in question, and
* You recompiled the libgcj verifier with -DVERIFY_DEBUG and sent me
  the verifier debug output for the method.  (FYI, this is likely to
  take a long time to get, since you'll get verifier output for all
  the methods that are loaded...)

Or you could send precise instructions for how to reproduce, including
URLs for the code and the like.  I can't guarantee that I can look at
it soon, so making a PR would be best in this case.

Tom


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]