[PATCH, RFC] Enable IBM long double for PPC32 Linux

[David Edelsohn]

>>>>> Jakub Jelinek writes:

Jakub> Issuing diagnostics when user does -msoft-float -mlong-double-128 is of
Jakub> course desirable.  But if it will be emitted many times during GCC
Jakub> bootstrap, some people might be confused and waste time looking into it.

	That is why inform() issues "note", not "warning".  This also
avoids problems compiling libgcc with -Werror.

	IMO, either we should issue a warning and use your specs hack or
issue a note and not use the hack.

	I am concerned about the hack because it depends on
-mlong-double-64 occurring last on the commandline.  If the inclusion
order of tmake Makefile frags causes -msoft-float to occur before
-mlong-double-128, the build will break.  This seems very fragile.

David