------- Comment #1 from kkojima at gcc dot gnu dot org 2006-07-03 22:49 -------
I've confirmed that 4.0 and 4.2 behave same as 4.1. So all compilers
which support -m4a option have this issue. It seems that the combine
is too smart and replaces the movua insn
(define_insn "movua"
[(set (match_operand:SI 0 "register_operand" "=z")
(sign_extract:SI (match_operand:SI 1 "unaligned_load_operand" "Sua>")
(const_int 32) (const_int 0)))]
"TARGET_SH4A_ARCH"
"movua.l %1,%0"
[(set_attr "type" "movua")])
with the usual load insn. I'm testing the patch below.
--- ORIG/trunk/gcc/config/sh/sh.md 2006-06-14 06:15:09.000000000 +0900
+++ LOCAL/trunk/gcc/config/sh/sh.md 2006-07-03 16:43:56.000000000 +0900
@@ -152,6 +152,7 @@
(UNSPEC_THUNK 36)
(UNSPEC_SP_SET 40)
(UNSPEC_SP_TEST 41)
+ (UNSPEC_MOVUA 42)
;; These are used with unspec_volatile.
(UNSPECV_BLOCKAGE 0)
@@ -11121,8 +11122,8 @@ mov.l\\t1f,r0\\n\\
(define_insn "movua"
[(set (match_operand:SI 0 "register_operand" "=z")
- (sign_extract:SI (match_operand:SI 1 "unaligned_load_operand" "Sua>")
- (const_int 32) (const_int 0)))]
+ (unspec:SI [(match_operand:SI 1 "unaligned_load_operand" "Sua>")]
+ UNSPEC_MOVUA))]
"TARGET_SH4A_ARCH"
"movua.l %1,%0"
[(set_attr "type" "movua")])
--
kkojima at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |wrong-code
Known to fail| |4.0.3 4.1.1 4.2.0
Summary|wrong movua.l instruction |Wrong aligned load is
|combine for sh4a |generated with -m4a on sh-*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28232
