X86: Make the movfp microop use FloatRegBits instead of FloatRegs.
authorGabe Black <gblack@eecs.umich.edu>
Wed, 5 Sep 2007 06:42:55 +0000 (23:42 -0700)
committerGabe Black <gblack@eecs.umich.edu>
Wed, 5 Sep 2007 06:42:55 +0000 (23:42 -0700)
commit6c689a3b4b3991342200bb2a433fdf7fbd76b4fc
treecff1b4cd2d41c5b6f4998fa3509d9177ca1b14a5
parent26ba7cc3b233ae2b3c036de2998e0c5e0a4c340d
X86: Make the movfp microop use FloatRegBits instead of FloatRegs.
This fixes a problem where interpreting arbitrary bits as floating point would
change what the value was. These values are legitimate because the fp
registers could be used to move around arbitrary data.

--HG--
extra : convert_revision : f1d7159ba137702b5722cc7c1b64ed7dd06c21de
src/arch/x86/isa/microops/regop.isa