[GCC][AARCH64] Canonicalize aarch64 widening simd plus insns

Committed on behalf of matthew.malcomson@arm.com

2018-07-24  Matthew Malcomson  <matthew.malcomson@arm.com>

	* config/aarch64/aarch64-simd.md
	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
	(aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
	(aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
	(aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
	(aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
	(aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
	(aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.

	* gcc.target/aarch64/vect-su-add-sub.c: New.

From-SVN: r262949

4 files changed, 124 insertions, 22 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index c7b921b..f1fdb1f 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,16 @@
+2018-07-24  Matthew Malcomson  <matthew.malcomson@arm.com>
+
+	* config/aarch64/aarch64-simd.md
+	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
+	(aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
+	(aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
+	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
+	(aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
+	(aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
+	(aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
+	(aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
+	(aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
+
 2018-07-24  Jakub Jelinek  <jakub@redhat.com>
 
 	PR middle-end/86627
diff --git a/gcc/config/aarch64/aarch64-simd.md b/gcc/config/aarch64/aarch64-simd.md
index f1784d7..9ee9bbe 100644
--- a/gcc/config/aarch64/aarch64-simd.md
+++ b/gcc/config/aarch64/aarch64-simd.md
@@ -3303,38 +3303,74 @@
   DONE;
 })
 
-(define_insn "aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>"
+(define_insn "aarch64_<ANY_EXTEND:su>subw<mode>"
   [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
-        (ADDSUB:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
-			(ANY_EXTEND:<VWIDE>
-			  (match_operand:VD_BHSI 2 "register_operand" "w"))))]
+	(minus:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
+	  (ANY_EXTEND:<VWIDE>
+	    (match_operand:VD_BHSI 2 "register_operand" "w"))))]
   "TARGET_SIMD"
-  "<ANY_EXTEND:su><ADDSUB:optab>w\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
-  [(set_attr "type" "neon_<ADDSUB:optab>_widen")]
+  "<ANY_EXTEND:su>subw\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
+  [(set_attr "type" "neon_sub_widen")]
 )
 
-(define_insn "aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal"
+(define_insn "aarch64_<ANY_EXTEND:su>subw<mode>_internal"
   [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
-        (ADDSUB:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
-			(ANY_EXTEND:<VWIDE>
-			  (vec_select:<VHALF>
-			   (match_operand:VQW 2 "register_operand" "w")
-			   (match_operand:VQW 3 "vect_par_cnst_lo_half" "")))))]
+	(minus:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
+	  (ANY_EXTEND:<VWIDE>
+	    (vec_select:<VHALF>
+	      (match_operand:VQW 2 "register_operand" "w")
+	      (match_operand:VQW 3 "vect_par_cnst_lo_half" "")))))]
   "TARGET_SIMD"
-  "<ANY_EXTEND:su><ADDSUB:optab>w\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vhalftype>"
-  [(set_attr "type" "neon_<ADDSUB:optab>_widen")]
+  "<ANY_EXTEND:su>subw\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vhalftype>"
+  [(set_attr "type" "neon_sub_widen")]
 )
 
-(define_insn "aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal"
+(define_insn "aarch64_<ANY_EXTEND:su>subw2<mode>_internal"
   [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
-        (ADDSUB:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
-			(ANY_EXTEND:<VWIDE>
-			  (vec_select:<VHALF>
-			   (match_operand:VQW 2 "register_operand" "w")
-			   (match_operand:VQW 3 "vect_par_cnst_hi_half" "")))))]
+	(minus:<VWIDE> (match_operand:<VWIDE> 1 "register_operand" "w")
+	  (ANY_EXTEND:<VWIDE>
+	    (vec_select:<VHALF>
+	      (match_operand:VQW 2 "register_operand" "w")
+	      (match_operand:VQW 3 "vect_par_cnst_hi_half" "")))))]
   "TARGET_SIMD"
-  "<ANY_EXTEND:su><ADDSUB:optab>w2\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
-  [(set_attr "type" "neon_<ADDSUB:optab>_widen")]
+  "<ANY_EXTEND:su>subw2\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
+  [(set_attr "type" "neon_sub_widen")]
+)
+
+(define_insn "aarch64_<ANY_EXTEND:su>addw<mode>"
+  [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
+	(plus:<VWIDE>
+	  (ANY_EXTEND:<VWIDE> (match_operand:VD_BHSI 2 "register_operand" "w"))
+	  (match_operand:<VWIDE> 1 "register_operand" "w")))]
+  "TARGET_SIMD"
+  "<ANY_EXTEND:su>addw\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
+  [(set_attr "type" "neon_add_widen")]
+)
+
+(define_insn "aarch64_<ANY_EXTEND:su>addw<mode>_internal"
+  [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
+	(plus:<VWIDE>
+	  (ANY_EXTEND:<VWIDE>
+	    (vec_select:<VHALF>
+	      (match_operand:VQW 2 "register_operand" "w")
+	      (match_operand:VQW 3 "vect_par_cnst_lo_half" "")))
+	  (match_operand:<VWIDE> 1 "register_operand" "w")))]
+  "TARGET_SIMD"
+  "<ANY_EXTEND:su>addw\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vhalftype>"
+  [(set_attr "type" "neon_add_widen")]
+)
+
+(define_insn "aarch64_<ANY_EXTEND:su>addw2<mode>_internal"
+  [(set (match_operand:<VWIDE> 0 "register_operand" "=w")
+	(plus:<VWIDE>
+	  (ANY_EXTEND:<VWIDE>
+	    (vec_select:<VHALF>
+	      (match_operand:VQW 2 "register_operand" "w")
+	      (match_operand:VQW 3 "vect_par_cnst_hi_half" "")))
+	  (match_operand:<VWIDE> 1 "register_operand" "w")))]
+  "TARGET_SIMD"
+  "<ANY_EXTEND:su>addw2\\t%0.<Vwtype>, %1.<Vwtype>, %2.<Vtype>"
+  [(set_attr "type" "neon_add_widen")]
 )
 
 (define_expand "aarch64_saddw2<mode>"
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index ca4d0bc..10488ca 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,7 @@
+2018-07-24  Matthew Malcomson  <matthew.malcomson@arm.com>
+
+	* gcc.target/aarch64/vect-su-add-sub.c: New.
+
 2018-07-24  Jakub Jelinek  <jakub@redhat.com>
 
 	PR middle-end/86627
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/vect_su_add_sub.c b/gcc/testsuite/gcc.target/aarch64/simd/vect_su_add_sub.c
new file mode 100644
index 0000000..338da54
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/simd/vect_su_add_sub.c
@@ -0,0 +1,49 @@
+/* { dg-do compile } */
+/* { dg-options "-O3" } */
+
+/* Ensure we use the signed/unsigned extend vectorized add and sub
+   instructions.  */
+#define N 1024
+
+int a[N];
+long c[N];
+long d[N];
+unsigned int ua[N];
+unsigned long uc[N];
+unsigned long ud[N];
+
+void
+add ()
+{
+  for (int i = 0; i < N; i++)
+    d[i] = a[i] + c[i];
+}
+/* { dg-final { scan-assembler-times "\[ \t\]saddw2\[ \t\]+" 1 } } */
+/* { dg-final { scan-assembler-times "\[ \t\]saddw\[ \t\]+" 1 } } */
+
+void
+subtract ()
+{
+  for (int i = 0; i < N; i++)
+    d[i] = c[i] - a[i];
+}
+/* { dg-final { scan-assembler-times "\[ \t\]ssubw2\[ \t\]+" 1 } } */
+/* { dg-final { scan-assembler-times "\[ \t\]ssubw\[ \t\]+" 1 } } */
+
+void
+uadd ()
+{
+  for (int i = 0; i < N; i++)
+    ud[i] = ua[i] + uc[i];
+}
+/* { dg-final { scan-assembler-times "\[ \t\]uaddw2\[ \t\]+" 1 } } */
+/* { dg-final { scan-assembler-times "\[ \t\]uaddw\[ \t\]+" 1 } } */
+
+void
+usubtract ()
+{
+  for (int i = 0; i < N; i++)
+    ud[i] = uc[i] - ua[i];
+}
+/* { dg-final { scan-assembler-times "\[ \t\]usubw2\[ \t\]+" 1 } } */
+/* { dg-final { scan-assembler-times "\[ \t\]usubw\[ \t\]+" 1 } } */