bootstrap/5191: libgcc2.h test to see if TImode can be defined is insufficient

[gcc@treblig.org]

>Number:         5191
>Category:       bootstrap
>Synopsis:       libgcc2.h test to see if TImode can be defined is insufficient
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Dec 25 16:56:00 PST 2001
>Closed-Date:
>Last-Modified:
>Originator:     Dr. David Alan Gilbert
>Release:        gcc-3.0.3
>Organization:
>Environment:
Host i686-linux (Debian/unstable)
(gcc-3.0.3 is clean ftp.gnu.org code)
>Description:
On an i686-linux host compiled for an alpha target we get an error that TI mode is unavailable while building libgcc2.h,
looking at it we see:

#if MIN_UNITS_PER_WORD > 4 && HOST_BITS_PER_WIDE_INT >= 64
/* These typedefs are usually forbidden on archs with UNITS_PER_WORD 4 */
typedef          int TItype     __attribute__ ((mode (TI)));
typedef unsigned int UTItype    __attribute__ ((mode (TI)));
#endif

giving the error:

In file included from ../../gcc-3.0.3/gcc/libgcc2.c:47:
../../gcc-3.0.3/gcc/libgcc2.h:101: no data type for mode `TI'
../../gcc-3.0.3/gcc/libgcc2.h:102: no data type for mode `TI'
make[2]: *** [libgcc/./_muldi3.o] Error 1



Now I guess that the test isn't strong enough; but I can't seem to find a definition in the info files of 
 HOST_BITS_PER_WIDE_INT
or indeed what a WIDE_INT corresponds to.
>How-To-Repeat:
(on an i686-linux host)
unpack gcc-3.0.3
configure for --target=alpha-linux

make
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: