The compiler is able to forward propagate a partial vector V4SF instruction
using XMM registers to a 3dNOW! V2SF instruction using MM registers. Prevent
unwanted transformation by tagging 3dNOW! V2SF instructions using generic
RTXes with "(unspec [(const_int 0)] UNSPEC_3DNOW)" tag.
PR target/117926
gcc/ChangeLog:
* config/i386/mmx.md (UNSPEC_3DNOW): New unspec.
(mmx_addv2sf3): Tag insn with UNSPEC_3DNOW tag.
(*mmx_addv2sf3): Ditto.
(mmx_sub2vsf3): Ditto.
(mmx_subrv2sf3): Ditto.
(*mmx_subv2sf3): Ditto.
(mmx_mulv2sf3): Ditto.
(mmx_<smaxmin:code>v2sf3): Ditto.
(*mmx_<smaxmin:code>v2sf3): Ditto.
(mmx_ieee_<ieee_maxmin>v2sf3): Ditto.
(mmx_eqv2sf3): Ditto.
(*mmx_eqv2sf3): Ditto.
(mmx_gtv2sf3): Ditto.
(mmx_gev2sf3): Ditto.
(mmx_fix_truncv2sfv2si2): Ditto.
(mmx_floatv2siv2sf2): Ditto.
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr117926.c: New test.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Uros.
diff --git a/gcc/config/i386/mmx.md b/gcc/config/i386/mmx.md
index 61a4f4d21ea..8a980ce069d 100644
--- a/gcc/config/i386/mmx.md
+++ b/gcc/config/i386/mmx.md
@@ -30,6 +30,8 @@
;; direction of the user via a builtin.
(define_c_enum "unspec" [
+ UNSPEC_3DNOW
+
UNSPEC_MOVNTQ
UNSPEC_PFRCP
UNSPEC_PFRCPIT1
@@ -727,17 +729,20 @@ (define_expand "<insn>v2sf3"
})
(define_expand "mmx_addv2sf3"
- [(set (match_operand:V2SF 0 "register_operand")
- (plus:V2SF
- (match_operand:V2SF 1 "nonimmediate_operand")
- (match_operand:V2SF 2 "nonimmediate_operand")))]
+ [(parallel
+ [(set (match_operand:V2SF 0 "register_operand")
+ (plus:V2SF
+ (match_operand:V2SF 1 "nonimmediate_operand")
+ (match_operand:V2SF 2 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
"TARGET_3DNOW"
"ix86_fixup_binary_operands_no_copy (PLUS, V2SFmode, operands);")
(define_insn "*mmx_addv2sf3"
[(set (match_operand:V2SF 0 "register_operand" "=y")
(plus:V2SF (match_operand:V2SF 1 "nonimmediate_operand" "%0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW && ix86_binary_operator_ok (PLUS, V2SFmode, operands)"
"pfadd\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxadd")
@@ -745,21 +750,26 @@ (define_insn "*mmx_addv2sf3"
(set_attr "mode" "V2SF")])
(define_expand "mmx_subv2sf3"
- [(set (match_operand:V2SF 0 "register_operand")
- (minus:V2SF (match_operand:V2SF 1 "register_operand")
- (match_operand:V2SF 2 "nonimmediate_operand")))]
+ [(parallel
+ [(set (match_operand:V2SF 0 "register_operand")
+ (minus:V2SF (match_operand:V2SF 1 "register_operand")
+ (match_operand:V2SF 2 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
"TARGET_3DNOW")
(define_expand "mmx_subrv2sf3"
- [(set (match_operand:V2SF 0 "register_operand")
- (minus:V2SF (match_operand:V2SF 2 "register_operand")
- (match_operand:V2SF 1 "nonimmediate_operand")))]
+ [(parallel
+ [(set (match_operand:V2SF 0 "register_operand")
+ (minus:V2SF (match_operand:V2SF 2 "register_operand")
+ (match_operand:V2SF 1 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
"TARGET_3DNOW")
(define_insn "*mmx_subv2sf3"
[(set (match_operand:V2SF 0 "register_operand" "=y,y")
(minus:V2SF (match_operand:V2SF 1 "nonimmediate_operand" "0,ym")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym,0")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym,0")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
"@
pfsub\t{%2, %0|%0, %2}
@@ -769,16 +779,19 @@ (define_insn "*mmx_subv2sf3"
(set_attr "mode" "V2SF")])
(define_expand "mmx_mulv2sf3"
- [(set (match_operand:V2SF 0 "register_operand")
- (mult:V2SF (match_operand:V2SF 1 "nonimmediate_operand")
- (match_operand:V2SF 2 "nonimmediate_operand")))]
+ [(parallel
+ [(set (match_operand:V2SF 0 "register_operand")
+ (mult:V2SF (match_operand:V2SF 1 "nonimmediate_operand")
+ (match_operand:V2SF 2 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
"TARGET_3DNOW"
"ix86_fixup_binary_operands_no_copy (MULT, V2SFmode, operands);")
(define_insn "*mmx_mulv2sf3"
[(set (match_operand:V2SF 0 "register_operand" "=y")
(mult:V2SF (match_operand:V2SF 1 "nonimmediate_operand" "%0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW && ix86_binary_operator_ok (MULT, V2SFmode, operands)"
"pfmul\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxmul")
@@ -827,10 +840,11 @@ (define_expand "<code>v2sf3"
})
(define_expand "mmx_<code>v2sf3"
- [(set (match_operand:V2SF 0 "register_operand")
- (smaxmin:V2SF
- (match_operand:V2SF 1 "nonimmediate_operand")
- (match_operand:V2SF 2 "nonimmediate_operand")))]
+ [(parallel
+ [(set (match_operand:V2SF 0 "register_operand")
+ (smaxmin:V2SF (match_operand:V2SF 1 "nonimmediate_operand")
+ (match_operand:V2SF 2 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
"TARGET_3DNOW"
{
if (!flag_finite_math_only || flag_signed_zeros)
@@ -851,9 +865,9 @@ (define_expand "mmx_<code>v2sf3"
(define_insn "*mmx_<code>v2sf3"
[(set (match_operand:V2SF 0 "register_operand" "=y")
- (smaxmin:V2SF
- (match_operand:V2SF 1 "nonimmediate_operand" "%0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (smaxmin:V2SF (match_operand:V2SF 1 "nonimmediate_operand" "%0")
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW && ix86_binary_operator_ok (<CODE>, V2SFmode, operands)"
"pf<maxmin_float>\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxadd")
@@ -871,7 +885,8 @@ (define_insn "mmx_ieee_<ieee_maxmin>v2sf3"
(unspec:V2SF
[(match_operand:V2SF 1 "register_operand" "0")
(match_operand:V2SF 2 "nonimmediate_operand" "ym")]
- IEEE_MAXMIN))]
+ IEEE_MAXMIN))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW"
"pf<ieee_maxmin>\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxadd")
@@ -1124,16 +1139,19 @@ (define_expand "vec_addsubv2sf3"
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(define_expand "mmx_eqv2sf3"
- [(set (match_operand:V2SI 0 "register_operand")
- (eq:V2SI (match_operand:V2SF 1 "nonimmediate_operand")
- (match_operand:V2SF 2 "nonimmediate_operand")))]
- "TARGET_3DNOW"
+ [(parallel
+ [(set (match_operand:V2SI 0 "register_operand")
+ (eq:V2SI (match_operand:V2SF 1 "nonimmediate_operand")
+ (match_operand:V2SF 2 "nonimmediate_operand")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)])]
+"TARGET_3DNOW"
"ix86_fixup_binary_operands_no_copy (EQ, V2SFmode, operands);")
(define_insn "*mmx_eqv2sf3"
[(set (match_operand:V2SI 0 "register_operand" "=y")
(eq:V2SI (match_operand:V2SF 1 "nonimmediate_operand" "%0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW && ix86_binary_operator_ok (EQ, V2SFmode, operands)"
"pfcmpeq\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxcmp")
@@ -1143,7 +1161,8 @@ (define_insn "*mmx_eqv2sf3"
(define_insn "mmx_gtv2sf3"
[(set (match_operand:V2SI 0 "register_operand" "=y")
(gt:V2SI (match_operand:V2SF 1 "register_operand" "0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW"
"pfcmpgt\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxcmp")
@@ -1153,7 +1172,8 @@ (define_insn "mmx_gtv2sf3"
(define_insn "mmx_gev2sf3"
[(set (match_operand:V2SI 0 "register_operand" "=y")
(ge:V2SI (match_operand:V2SF 1 "register_operand" "0")
- (match_operand:V2SF 2 "nonimmediate_operand" "ym")))]
+ (match_operand:V2SF 2 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW"
"pfcmpge\t{%2, %0|%0, %2}"
[(set_attr "type" "mmxcmp")
@@ -1485,7 +1505,8 @@ (define_expand "fixuns_truncv2sfv2si2"
(define_insn "mmx_fix_truncv2sfv2si2"
[(set (match_operand:V2SI 0 "register_operand" "=y")
- (fix:V2SI (match_operand:V2SF 1 "nonimmediate_operand" "ym")))]
+ (fix:V2SI (match_operand:V2SF 1 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW"
"pf2id\t{%1, %0|%0, %1}"
[(set_attr "type" "mmxcvt")
@@ -1526,7 +1547,8 @@ (define_expand "floatunsv2siv2sf2"
(define_insn "mmx_floatv2siv2sf2"
[(set (match_operand:V2SF 0 "register_operand" "=y")
- (float:V2SF (match_operand:V2SI 1 "nonimmediate_operand" "ym")))]
+ (float:V2SF (match_operand:V2SI 1 "nonimmediate_operand" "ym")))
+ (unspec [(const_int 0)] UNSPEC_3DNOW)]
"TARGET_3DNOW"
"pi2fd\t{%1, %0|%0, %1}"
[(set_attr "type" "mmxcvt")
diff --git a/gcc/testsuite/gcc.target/i386/pr117926.c
b/gcc/testsuite/gcc.target/i386/pr117926.c
new file mode 100644
index 00000000000..6b7aca3726d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr117926.c
@@ -0,0 +1,18 @@
+/* PR target/117926 */
+/* { dg-do compile { target { ! ia32 } } } */
+/* { dg-options "-O2 -m3dnow" } */
+
+struct s
+{
+ int i[2];
+ float f[2];
+ double d;
+};
+
+void f (struct s *s)
+{
+ s->f[0] = s->i[0];
+ s->f[1] = s->i[1];
+}
+
+/* { dg-final { scan-assembler-not "pi2fd" } } */
