[Bug libgcc/93192] New: [m68k] incorrect conversion of inf and nan in __truncxfdf2

[admin@tho-otto.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93192

            Bug ID: 93192
           Summary: [m68k] incorrect conversion of inf and nan in
                    __truncxfdf2
           Product: gcc
           Version: 9.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libgcc
          Assignee: unassigned at gcc dot gnu.org
          Reporter: admin@tho-otto.de
  Target Milestone: ---

Created attachment 47606
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=47606&action=edit
Suggested patch

The implementation of _truncxfdf2 (long double to double) incorrectly converts
INFs and NaNs to very large numbers (8.98847e307).

Attached is a suggested patch. It also fixes the test case for a zero exponent,
which in the current version also tests the empty 16 bits of the long double
format for being zero, but should ignore them.