[jboss-jira] [JBoss JIRA] Created: (JASSIST-71) Array access replacement generates invalid bytecode when padding is required

[Jason T. Greene (JIRA)]

Array access replacement generates invalid bytecode when padding is required
----------------------------------------------------------------------------

                 Key: JASSIST-71
                 URL: https://jira.jboss.org/jira/browse/JASSIST-71
             Project: Javassist
          Issue Type: Bug
            Reporter: Jason T. Greene
            Assignee: Shigeru Chiba


When an array access instruction is replaced before a switch instruction, and padding is added to align the switch, the array instruction remains, which is invalid:

public void test();
  Code:
   Stack=2, Locals=2, Args_size=1
   0:	aconst_null
   1:	astore_1
   2:	invokestatic	#17; //Method $SWITCH_TABLE$RULE_TYPE:()[I
   5:	aload_1
   6:	invokevirtual	#20; //Method RULE_TYPE.ordinal:()I
   9:	invokestatic	#67; //Method arrayReadInt:(Ljava/lang/Object;I)I
   12:	nop 
   13:	iaload                               <=========================== INVALID!!!!!!!!!!!! 
   14:	tableswitch{ //1 to 1
		1: 32;
		default: 32 }
   32:	return

This is typically caught by the bytecode verifier

 java.lang.VerifyError: (class: com/lm/dataModel/rules/StatusRule, method: propagateFromHereOnUpTree signature: (Lcom/lm/dataModel/ANode;Lcom/lm/dataModel/ANode$Source;Lcom/lm/dataModel/ANode$StatusValue;)Z) Unable to pop operand off an empty stack
     at com.lm.dataModel.ANode.setStatusValueWithPropagation(ANode.java:301)
     at com.lm.seem.StatusTableModel.setValueAt(StatusTableModel.java:133)

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira