IV68043: INCORRECT VALUE PRODUCED FROM FLOATING POINT COMPUTATION

APAR status

• Closed as program error.

Error description

• Error Message: A Java program behaves incorrectly or produces
  incorrect output because a floating point value is incorrect
  during program execution. The incorrect value will be a
  different valid floating point value that is either an
  intermediate result of a floating point calculation or the value
  of floating point variable somewhere in the program.
  .
  Stack Trace: N/A
  .
  The Java method that generates the invalid value will contain at
  least one lookupswitch bytecode.
  

Local fix

• The problem can be worked around using the -Xjit:exclude=
  command line option to exclude the method that generates the
  faulty floating point value from compilation, this is likely to
  cause some performance degradation.
  

Problem summary

• The invalid value results from floating point value dependency
  information not being fully propagated across lookupswitch
  bytecodes during JIT compilation of the method.
  

Problem conclusion

• The JIT compiler has been enhanced to ensure floating point
  value dependency information is propagated across lookupswitch
  bytecodes in all cases to ensure values are correctly computed.
  .
  This APAR will be fixed in the following Java Releases:
     7 R1 SR3       (7.1.3.0)
     7    SR9       (7.0.9.0)
     6 R1 SR8 FP4   (6.1.8.4)
     6    SR16 FP4  (6.0.16.4)
  .
  Contact your IBM Product's Service Team for these Service
  Refreshes and Fix Packs.
  For those running stand-alone, information about the Service
  Refreshes and Fix Packs can be found at:
             https://www.ibm.com/developerworks/java/jdk/
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  JIT

• Fixed component ID

  620700124

Applicable component levels

• R130 PSY

     UP

• R260 PSY

     UP

• R600 PSY

     UP