This is the mail archive of the libc-hacker@cygnus.com mailing list for the glibc project.


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

libm test failures, current CVS



For all three sizes of floating-point, inline or not, on x86:

    Fail: fdim (9, 0) = 9
    Result:
    is:       -9.00000000000000000000e+00  -0x1.20000000000000000000p+3
    should be:   9.00000000000000000000e+00   0x1.20000000000000000000p+3
    difference:  1.80000000000000000000e+01   0x1.20000000000000000000p+4
    Fail: fdim (0, -9) = 9
    Result:
    is:       -9.00000000000000000000e+00  -0x1.20000000000000000000p+3
    should be:   9.00000000000000000000e+00   0x1.20000000000000000000p+3
    difference:  1.80000000000000000000e+01   0x1.20000000000000000000p+4
    Fail: fdim (+inf, 9) = +inf
    Value: -inf  -inf
    Fail: fdim (+inf, -9) = +inf
    Value: -inf  -inf
    Fail: fdim (+9, -inf) = +inf
    Value: -inf  -inf
    Fail: fdim (-9, -inf) = +inf
    Value: -inf  -inf

presumably provoked by this change:

        * sysdeps/libm-i387/s_fdim.S: Change fsubp to fsubrp for picky gas.
        * sysdeps/libm-i387/s_fdimf.S: Likewise.
        * sysdeps/libm-i387/s_fdiml.S: Likewise.

I don't know how to fix it.

zw


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