https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90530
John David Anglin <danglin at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |rguenth at gcc dot gnu.org
Summary|[10 Regression] Invalid |[9,10 Regression] Invalid
|SUBREG insn generated by |SUBREG insn generated by
|reload |reload
--- Comment #5 from John David Anglin <danglin at gcc dot gnu.org> ---
Problematic change was back ported to gcc-9.
I'm testing a change to reject SUBREG operands in the mult patterns.
They require operands be loaded in floating-point registers. However,
it seems to that this problem might occur in any pattern that requires
a float register in integer mode given we can't allow mode changes.
So, I'm thinking is this is just a work around.
