Home • About Us • A-Z Index • Search * Contact Us • Register • Login • Press • Shop

The Open Brand -- Problem Reporting and Interpretations System

Problem Report 0383 Details

Help Show help | Quick Search | Submit a Test Suite Support Request | Click here to view your privileges

This page provides all information on Problem Report 0383.

Report 0383 Actions

    Problem Report Number 0383
    Submitter's Classification Test Suite problem
    State Resolved
    Resolution Rejected (REJ)
    Problem Resolution ID REJ.X.0002
    Raised 1992-09-18 08:00
    Updated 2003-03-13 08:00
    Published null
    Product Standard Internationalised System Calls and Libraries (XPG4)
    Certification Program The Open Brand certification program
    Test Suite VSX4 version 4.2.3
    Test Identification ANSI.os/maths/exp 1
    Problem Summary PG4R.002 This problem also affects pow{1} for the same reason. The problem occurs because our hardware recognises floating point underflow. To quote from our hardware manual: "When a result cannot be expressed...
    Problem Text

    This problem also affects pow{1} for the same reason.

    The problem occurs because our hardware recognises floating
    point underflow. To quote from our hardware manual:

    "When a result cannot be expressed in normalised form with
    a true exponent greater than -65, that is, it would require
    a negative characteristic, underflow occurs and the result
    is made true zero, and unless the condition is masked an
    interrupt occurs"

    Our system runs with all arithmetic exception conditions
    enabled, and therefore when the error occurs and interrupt
    is raised. This interrupt is translated into SIGFPE.

    The underflow occurs in the test code that subtracts the result
    obtained from the library function being tested from the one
    obtained from the suites own version. In the cases where
    overflow occurs, the values returned by both functions are
    almost identical and have an exponent equal to that of the
    smallest number it is possible to represent as a normalised
    number.

    The subtraction results in a denormalised number which causes
    the underflow condition to occur.

    One way to fix the suite could be to introduce user routines
    in userintf.c which could be called to turn on or off the
    exception conditions as a software solution to the problem
    of underflow is not obvious.
    Test Output

    /tset/ANSI.os/maths/exp/T.exp 1 Unresolved

    child process was terminated by signal 8 (SIGFPE)
    *************************************************************

    Review Information

    Review Type TSMA Review
    Start Date null
    Completed null
    Status Complete
    Review Recommendation No Resolution Given
    Review Response
    The fact that the SIGFPE signal is not being ignored is something of a surprise.
    It is normal to ignore this signal at the start of each of the maths function
    testsets. Have there been any code changes that alter the behaviour of VSX?

    It is also noted that the waiver request only cites exp test 1, though the
    commentary also mentions pow test 1. Should there be a waiver request for
    pow test 1 also?

    Review Type SA Review
    Start Date null
    Completed null
    Status Complete
    Review Resolution Rejected (REJ)
    Review Conclusion
    This waiver request is refused. We suggest that an error report be submitted.

    Problem Reporting System Options:

     

    Back   

Contact the Certification Authority