[Bug math/15491] New: x86 nearbyint implementations wrongly clear all exceptions

["jsm28 at gcc dot gnu.org" <sourceware-bugzilla at sourceware dot org>]

http://sourceware.org/bugzilla/show_bug.cgi?id=15491

             Bug #: 15491
           Summary: x86 nearbyint implementations wrongly clear all
                    exceptions
           Product: glibc
           Version: 2.17
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
        AssignedTo: unassigned@sourceware.org
        ReportedBy: jsm28@gcc.gnu.org
    Classification: Unclassified


The implementations of nearbyint functions using x87 floating point (x86 all
versions, x86_64 long double only) use the fclex instruction, which clears any
exceptions that were raised before the function was called.  These functions
must not clear exceptions that were raised before they were called.

I think fixing this will require saving and restoring the whole floating-point
environment as the way of avoiding raising "inexact" (which is what fesetenv
does, so use of frndint should still be faster than the C versions) - except
that if frndint raises an "invalid" exception for a sNaN operand, that *should*
be added into the restored environment.

-- 
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.