[Bug math/15442] New: MIPS NewABI sqrtl (soft-fp) test-ldouble sqrt (qNaN) failure

["macro at linux-mips dot org" <sourceware-bugzilla at sourceware dot org>]

http://sourceware.org/bugzilla/show_bug.cgi?id=15442

             Bug #: 15442
           Summary: MIPS NewABI sqrtl (soft-fp) test-ldouble sqrt (qNaN)
                    failure
           Product: glibc
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
        AssignedTo: macro@linux-mips.org
        ReportedBy: macro@linux-mips.org
                CC: tschwinge@sourceware.org
    Classification: Unclassified
            Target: mips-linux-gnu


This problem:

Failure: sqrt (qNaN) == qNaN: Exception "Invalid operation" set
Failure: Test: sqrt (qNaN) == qNaN
Result:
 is:         sNaN
 should be:  qNaN

is reported for sqrtl by test-ldouble for the MIPS/Linux target and the
n64 and n64 ABIs.  This is because soft-fp does not respect the MIPS NaN
encoding convention, where the quiet bit being zero denotes a qNaN.
This is the only soft-fp function used by the MIPS port.

There are other failures too, for negative numeric inputs, e.g. -1, but
in those cases the result is passed through __kernel_standard that
returns a double value rather than a long double value.  Therefore the
result is converted with __extenddftf2, that is a part of GCC's fp-bit
soft-float support.  That support suffers from a related bug being
discussed here:

http://gcc.gnu.org/ml/gcc-patches/2013-04/msg00278.html

and therefore the remaining failures cannot be removed by simply
correcting our soft-fp support.

A proposed fix has been posted here:

http://sourceware.org/ml/libc-alpha/2013-04/msg00646.html

-- that removes the test-ldouble failure referred to above.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.