What's the matter with paradoxical subregs of memory? So, as far
as I understand, Recog refuses to match them unless they are
explicitly expressed as a pattern. On the other hand, Combine
does not know that they are "outlawed" and happily generates
them.
This happens sometimes with some new patterns for s390x for the
RISBG instruction, which does complex operations on registers. As
it's beneficial to allow memory operands before reload,
paradoxical subregs of memory are sometimes combined from such
patterns, for example
(set (reg:DI 68)
(and:DI (subreg:DI (lshiftrt:SI (reg:SI 67 [ *f_5(D) ])
(const_int 8 [0x8])) 0)
(const_int 16777215 [0xffffff])))
+
(set (reg/i:DI 2 %r2)
(and:DI (reg:DI 68)
(const_int 1 [0x1])))
=
(set (reg/i:DI 2 %r2)
(and:DI (subreg:DI (mem:QI (plus:DI (reg:DI 2 %r2 [ f ])
^^^^^^^^^^^^^^^^^
(const_int 2 [0x2])) [1 *f_5(D)+2 S1 A16]) 0)
(const_int 1 [0x1])))
Which doesn't match because of the paradoxical subreg of memory (I
think).
Wouldn't it be better to allow that inside combine, then ann a
pass right after it that forces a reaload into a pseudo for such
memory references (for the targets that don't want them)? (Not
sure if theat would result in a noticeable gain or not.)
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
