This patch implements the [u]avgM3_floor and [u]avgM3_ceil optabs for SVE2.

From-SVN: r271739

gcc/ChangeLog

+2019-05-29  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
+
+	* config/aarch64/aarch64-c.c: Added TARGET_SVE2.
+	* config/aarch64/aarch64-sve2.md: New file.
+	(<u>avg<mode>3_floor): New pattern.
+	(<u>avg<mode>3_ceil): Likewise.
+	(*<sur>h<addsub><mode>): Likewise.
+	* config/aarch64/aarch64.h: Added AARCH64_ISA_SVE2 and TARGET_SVE2.
+	* config/aarch64/aarch64.md: Include aarch64-sve2.md.
+
 2019-05-29  Jakub Jelinek  <jakub@redhat.com>
 
 	PR bootstrap/90543

gcc/config/aarch64/aarch64-c.c

@@ -146,6 +146,7 @@ aarch64_update_cpp_builtins (cpp_reader *pfile)
 	bits = 0;
       builtin_define_with_int_value ("__ARM_FEATURE_SVE_BITS", bits);
     }
+  aarch64_def_or_undef (TARGET_SVE2, "__ARM_FEATURE_SVE2", pfile);
 
   aarch64_def_or_undef (TARGET_LSE, "__ARM_FEATURE_ATOMICS", pfile);
   aarch64_def_or_undef (TARGET_AES, "__ARM_FEATURE_AES", pfile);

gcc/config/aarch64/aarch64-sve2.md

+;; Machine description for AArch64 SVE2.
+;; Copyright (C) 2019 Free Software Foundation, Inc.
+;; Contributed by ARM Ltd.
+;;
+;; This file is part of GCC.
+;;
+;; GCC is free software; you can redistribute it and/or modify it
+;; under the terms of the GNU General Public License as published by
+;; the Free Software Foundation; either version 3, or (at your option)
+;; any later version.
+;;
+;; GCC is distributed in the hope that it will be useful, but
+;; WITHOUT ANY WARRANTY; without even the implied warranty of
+;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+;; General Public License for more details.
+;;
+;; You should have received a copy of the GNU General Public License
+;; along with GCC; see the file COPYING3.  If not see
+;; <http://www.gnu.org/licenses/>.
+
+;; Integer average (floor).
+(define_expand "<u>avg<mode>3_floor"
+  [(set (match_operand:SVE_I 0 "register_operand")
+	(unspec:SVE_I
+	  [(match_dup 3)
+	   (unspec:SVE_I [(match_operand:SVE_I 1 "register_operand")
+			  (match_operand:SVE_I 2 "register_operand")]
+			 HADD)]
+	  UNSPEC_MERGE_PTRUE))]
+  "TARGET_SVE2"
+  {
+    operands[3] = force_reg (<VPRED>mode, CONSTM1_RTX (<VPRED>mode));
+  }
+)
+
+;; Integer average (rounding).
+(define_expand "<u>avg<mode>3_ceil"
+  [(set (match_operand:SVE_I 0 "register_operand")
+	(unspec:SVE_I
+	  [(match_dup 3)
+	   (unspec:SVE_I [(match_operand:SVE_I 1 "register_operand")
+			  (match_operand:SVE_I 2 "register_operand")]
+			 RHADD)]
+	  UNSPEC_MERGE_PTRUE))]
+  "TARGET_SVE2"
+  {
+    operands[3] = force_reg (<VPRED>mode, CONSTM1_RTX (<VPRED>mode));
+  }
+)
+
+;; Predicated halving addsub.
+(define_insn "*<sur>h<addsub><mode>"
+  [(set (match_operand:SVE_I 0 "register_operand" "=w, ?&w")
+	(unspec:SVE_I
+	  [(match_operand:<VPRED> 1 "register_operand" "Upl, Upl")
+	   (unspec:SVE_I [(match_operand:SVE_I 2 "register_operand" "%0, w")
+			  (match_operand:SVE_I 3 "register_operand" "w, w")]
+			 HADDSUB)]
+	  UNSPEC_MERGE_PTRUE))]
+  "TARGET_SVE2"
+  "@
+   <sur>h<addsub>\t%0.<Vetype>, %1/m, %0.<Vetype>, %3.<Vetype>
+   movprfx\t%0, %2\;<sur>h<addsub>\t%0.<Vetype>, %1/m, %0.<Vetype>, %3.<Vetype>"
+  [(set_attr "movprfx" "*,yes")]
+)
\ No newline at end of file

gcc/config/aarch64/aarch64.h

@@ -232,6 +232,7 @@ extern unsigned aarch64_architecture_version;
 #define AARCH64_ISA_V8_2	   (aarch64_isa_flags & AARCH64_FL_V8_2)
 #define AARCH64_ISA_F16		   (aarch64_isa_flags & AARCH64_FL_F16)
 #define AARCH64_ISA_SVE            (aarch64_isa_flags & AARCH64_FL_SVE)
+#define AARCH64_ISA_SVE2	   (aarch64_isa_flags & AARCH64_FL_SVE2)
 #define AARCH64_ISA_V8_3	   (aarch64_isa_flags & AARCH64_FL_V8_3)
 #define AARCH64_ISA_DOTPROD	   (aarch64_isa_flags & AARCH64_FL_DOTPROD)
 #define AARCH64_ISA_AES	           (aarch64_isa_flags & AARCH64_FL_AES)
@@ -277,6 +278,9 @@ extern unsigned aarch64_architecture_version;
 /* SVE instructions, enabled through +sve.  */
 #define TARGET_SVE (AARCH64_ISA_SVE)
 
+/* SVE2 instructions, enabled through +sve2.  */
+#define TARGET_SVE2 (AARCH64_ISA_SVE2)
+
 /* ARMv8.3-A features.  */
 #define TARGET_ARMV8_3	(AARCH64_ISA_V8_3)
 

gcc/config/aarch64/aarch64.md

@@ -7244,3 +7244,6 @@
 
 ;; SVE.
 (include "aarch64-sve.md")
+
+;; SVE2.
+(include "aarch64-sve2.md")

gcc/testsuite/ChangeLog

+2019-05-29  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
+
+	* gcc.target/aarch64/sve2/aarch64-sve2.exp: New file, regression driver
+	for AArch64 SVE2.
+	* gcc.target/aarch64/sve2/average_1.c: New test.
+	* lib/target-supports.exp (check_effective_target_aarch64_sve2): New
+	helper.
+	(check_effective_target_aarch64_sve1_only): Likewise.
+	(check_effective_target_aarch64_sve2_hw): Likewise.
+	(check_effective_target_vect_avg_qi): Check for SVE1 only.
+
 2019-05-29  Sam Tebbs  <sam.tebbs@arm.com>
 
 	* gcc.target/aarch64/return_address_sign_b_1.c: New file.

gcc/testsuite/gcc.target/aarch64/sve2/aarch64-sve2.exp

+#  Specific regression driver for AArch64 SVE2.
+#  Copyright (C) 2009-2019 Free Software Foundation, Inc.
+#  Contributed by ARM Ltd.
+#
+#  This file is part of GCC.
+#
+#  GCC is free software; you can redistribute it and/or modify it
+#  under the terms of the GNU General Public License as published by
+#  the Free Software Foundation; either version 3, or (at your option)
+#  any later version.
+#
+#  GCC is distributed in the hope that it will be useful, but
+#  WITHOUT ANY WARRANTY; without even the implied warranty of
+#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+#  General Public License for more details.
+#
+#  You should have received a copy of the GNU General Public License
+#  along with GCC; see the file COPYING3.  If not see
+#  <http://www.gnu.org/licenses/>.  */
+
+# GCC testsuite that uses the `dg.exp' driver.
+
+# Exit immediately if this isn't an AArch64 target.
+if {![istarget aarch64*-*-*] } then {
+  return
+}
+
+# Load support procs.
+load_lib gcc-dg.exp
+
+# If a testcase doesn't have special options, use these.
+global DEFAULT_CFLAGS
+if ![info exists DEFAULT_CFLAGS] then {
+    set DEFAULT_CFLAGS " -ansi -pedantic-errors"
+}
+
+# Initialize `dg'.
+dg-init
+
+# Force SVE2 if we're not testing it already.
+if { [check_effective_target_aarch64_sve2] } {
+    set sve2_flags ""
+} else {
+    set sve2_flags "-march=armv8.5-a+sve2"
+}
+
+# Main loop.
+dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/*.\[cCS\]]] \
+    $sve2_flags $DEFAULT_CFLAGS
+
+# All done.
+dg-finish

gcc/testsuite/gcc.target/aarch64/sve2/average_1.c

+/* { dg-do compile } */
+/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */
+
+#include <stdint.h>
+
+#define AVERAGE(TYPE, BIGGER, RND)					  \
+TYPE __attribute__ ((noinline, noclone))				  \
+avg_##TYPE##_##RND (TYPE *restrict x, TYPE *restrict y, TYPE *restrict z, \
+		    int n)						  \
+{									  \
+  for (int i = 0; i < n; i++)						  \
+    {									  \
+      z[i] = ((BIGGER)x[i] + y[i] + RND) >> 1;				  \
+    }									  \
+}
+
+AVERAGE (int8_t, int64_t, 0)
+AVERAGE (int16_t, int64_t, 0)
+AVERAGE (int32_t, int64_t, 0)
+AVERAGE (uint8_t, uint64_t, 0)
+AVERAGE (uint16_t, uint64_t, 0)
+AVERAGE (uint32_t, uint64_t, 0)
+AVERAGE (int8_t, int64_t, 1)
+AVERAGE (int16_t, int64_t, 1)
+AVERAGE (int32_t, int64_t, 1)
+AVERAGE (uint8_t, uint64_t, 1)
+AVERAGE (uint16_t, uint64_t, 1)
+AVERAGE (uint32_t, uint64_t, 1)
+
+/* { dg-final { scan-tree-dump-times "vectorized 1 loops in function" 12 "vect" } } */
+
+/* { dg-final { scan-assembler-times {\tshadd\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tshadd\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tshadd\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
+
+/* { dg-final { scan-assembler-times {\tuhadd\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tuhadd\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tuhadd\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
+
+/* { dg-final { scan-assembler-times {\tsrhadd\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tsrhadd\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
+/* { dg-final { scan-assembler-times {\tsrhadd\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
+
+/* { dg-final { scan-assembler-times {\turhadd\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
+/* { dg-final { scan-assembler-times {\turhadd\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
+/* { dg-final { scan-assembler-times {\turhadd\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */

gcc/testsuite/lib/target-supports.exp

@@ -3306,6 +3306,24 @@ proc check_effective_target_aarch64_sve { } {
     }]
 }
 
+# Return 1 if this is an AArch64 target supporting SVE2.
+proc check_effective_target_aarch64_sve2 { } {
+    if { ![istarget aarch64*-*-*] } {
+	return 0
+    }
+    return [check_no_compiler_messages aarch64_sve2 assembly {
+	#if !defined (__ARM_FEATURE_SVE2)
+	#error FOO
+	#endif
+    }]
+}
+
+# Return 1 if this is an AArch64 target only supporting SVE (not SVE2).
+proc check_effective_target_aarch64_sve1_only { } {
+    return [expr { [check_effective_target_aarch64_sve]
+		   && ![check_effective_target_aarch64_sve2] }]
+}
+
 # Return the size in bits of an SVE vector, or 0 if the size is variable.
 proc aarch64_sve_bits { } {
     return [check_cached_effective_target aarch64_sve_bits {
@@ -4326,6 +4344,22 @@ proc check_effective_target_aarch64_sve_hw { } {
     }]
 }
 
+# Return true if this is an AArch64 target that can run SVE2 code.
+
+proc check_effective_target_aarch64_sve2_hw { } {
+    if { ![istarget aarch64*-*-*] } {
+	return 0
+    }
+    return [check_runtime aarch64_sve2_hw_available {
+	int
+	main (void)
+	{
+	  asm volatile ("addp z0.b, p0/m, z0.b, z1.b");
+	  return 0;
+	}
+    }]
+}
+
 # Return true if this is an AArch64 target that can run SVE code and
 # if its SVE vectors have exactly BITS bits.
 
@@ -6063,7 +6097,7 @@ proc check_effective_target_vect_usad_char { } {
 
 proc check_effective_target_vect_avg_qi {} {
     return [expr { [istarget aarch64*-*-*]
-		   && ![check_effective_target_aarch64_sve] }]
+		   && ![check_effective_target_aarch64_sve1_only] }]
 }
 
 # Return 1 if the target plus current options supports a vector