Re: IllegalStateException: Try catch blocks must be visited before their labels

Hi Rajaraman,

it is difficult because you don't provide a proper example.

Currently, it is unclear, which transformations your system does before
your CheckClassAdaptor and which ones after - and it is totally unclear
whether your transformation holds the verification criteria in between.

Only If we know all this facts, we can conclude about a concrete problem
an whether output of CheckClassAdaptor gives the right hint.

Best regards

Tronje

Rajaraman Santhanam wrote:
> Hi,
> 
> Can anybody comment on this? 
> 
> 
> Thanks.
> 
> ---- On Mon, 31 Jan 2011 19:43:31 +0530 Rajaraman Santhanam <[email protected]> wrote ---- 
> 
> Hi Tronje,
> 
> What you say makes sense. Either VM is a bit lenient or CheckClassAdaptor is strict.
> 
> 
> I tried with a simple java file and this seems to be the behavior. I get the issue only when I keep CheckClassAdapter in front of my visitors.
> MyClassAdapter adapter =  new MyClassAdapter(new CheckClassAdapter(writer), className, classBeingRedefined); 
> 
> 
> 
> Also I found that, if I run the CheckClassAdapter(as below) after bytecode transformation I don't see any issue at all.
> StringWriter checkClassSW = new StringWriter();
> PrintWriter checkClassPW = new PrintWriter(checkClassSW);
> CheckClassAdapter.verify(reader, loader, dump, checkClassPW);
> System.out.println(checkClassSW.toString());
> try
> {
>             checkClassPW.close();
>             checkClassSW.close();
> }
> catch(Exception ex)
> {
>             ex.printStackTrace();
> } 
> 
> 
> It looks like CheckClassAdaptor looks for some state and it finds visitLabel called before trycatchblock. However the same adaptor behaves differently :-( 
> 
> Also attaching the sources I used for testing.
> 
> 
> Thanks,
> Rajaraman
> 
> 
> 
> 
> ---- On Sat, 29 Jan 2011 21:15:52 +0530 Tronje Krop <[email protected]> wrote ---- 
> 
> Hi Rajaraman, 
>  
> it is as the error says. You have instrumented the byte-code in a way 
> that at least one of the labels has been visited before defining the 
> try-catch-block. The CheckMethodAdapter checks this requirement and 
> throws the exception. 
>  
> If you are right with your claim, it seems that the VM-Verifier has a 
> relaxed condition which allows to define the try-catch-block using labels 
> have being visited before. 
>  
> I didn't find a rule in the VM-specification that prohibits this, but 
> I have not looked very close. Therefore, I think the CheckClassAdapter 
> may be to restrictive unless ASM has a problem with this relaxed 
> condition. 
>  
> I think we must investigate this issue closer. 
>  
> CU Tronje 
>  
>  
> Rajaraman Santhanam wrote: 
> > Hi, 
> > 
> > I get this exception if I put CheckClassAdapter in the visitors chain before ClassWriter. The code I use is below: 
> > ClassReader reader = new ClassReader(classfileBuffer); 
> > ClassWriter writer = new ClassWriter(reader, ClassWriter.COMPUTE_MAXS); 
> > 
> > 
> > BasicClassAdapter adapter = new BasicClassAdapter(new CheckClassAdapter(writer), className, classBeingRedefined); 
> > reader.accept(adapter, ClassReader.SKIP_DEBUG); 
> > 
> > By running this I get error with details: 
> > java.lang.IllegalStateException: Try catch blocks must be visited before their labels 
> > at org.objectweb.asm.util.CheckMethodAdapter.visitTryCatchBlock(CheckMethodAdapter.java:679) 
> > at org.objectweb.asm.MethodAdapter.visitTryCatchBlock(MethodAdapter.java:170) 
> > at test.agent.instrumentation.BasicMethodAdapter.visitMaxs(BasicMethodAdapter.java:86) 
> > 
> > However my instrumentation code work perfectly well if run this without CheckClassAdapter. Not sure what I'm missing here. 
> > 
> > 
> > I tried this when seeing the mail from Eugene to use CheckClassAdapter to catch various issues. 
> > 
> > 
> > My BasicClassAdapter is the implementation of FinallyAdapter from asm-transformations.pdf 
> > 
> > 
> > Can somebody help me out here? 
> > 
> > 
> > Regards, 
> > Rajaraman 
> > 
> > 
> > 
> > 
>  
> 

-- 
     Tronje Krop <[email protected]>

   Encrypted eMail welcome! GPG/PGP-Key: 0x9AD43A05
  68E5 A3D3 75A0 B096 AC75  62D5 8EEE 3D18 9AD4 3A05

 No responsibility is taken for the correctness of the
previous information. Please delete if you receive this
                  mail unintentional.

-- 
You receive this message as a subscriber of the [email protected] mailing list.
To unsubscribe: mailto:[email protected]
For general help: mailto:[email protected]?subject=help
OW2 mailing lists service home page: http://www.ow2.org/wws