Sync with svn rev. 192135

Signed-off-by: Bernhard Rosenkraenzer <Bernhard.Rosenkranzer@linaro.org>

3 files changed, 58 insertions, 101 deletions

diff --git a/gcc/ChangeLog.aarch64 b/gcc/ChangeLog.aarch64
index 3ac47220f..34bdc16f6 100644
--- a/gcc/ChangeLog.aarch64
+++ b/gcc/ChangeLog.aarch64
@@ -1,3 +1,23 @@
+2012-10-05  Tejas Belagod  <tejas.belagod@arm.com>
+
+	* config/aarch64/arm_neon.h (vfma_n_f32, vfmaq_n_f32, vfmaq_n_f64): New.
+
+2012-10-05  Tejas Belagod  <tejas.belagod@arm.com>
+
+	* config/aarch64/arm_neon.h (vqmlalh_lane_s16, vqmlalh_s16,
+	vqmlals_lane_s32, vqmlals_s32, vqmlslh_lane_s16, vqmlslh_s16,
+	vqmlsls_lane_s32, vqmlsls_s32): Remove old temporary inline asm 
+	implementations.
+
+2012-10-05  Tejas Belagod  <tejas.belagod@arm.com>
+
+	* config/aarch64/arm_neon.h (vbslq_f64): Fix parameter type.
+
+2012-10-05  Sofiane Naci  <sofiane.naci@arm.com>
+
+	* config/aarch64/aarch64.md (*fnmadd<mode>4): Add missing
+	constraints.
+
 2012-10-01  Ian Bolton  <ian.bolton@arm.com>
 	    Richard Henderson  <rth@redhat.com>
 
diff --git a/gcc/config/aarch64/aarch64.md b/gcc/config/aarch64/aarch64.md
index 241bc9213..1df4d02c2 100644
--- a/gcc/config/aarch64/aarch64.md
+++ b/gcc/config/aarch64/aarch64.md
@@ -2516,10 +2516,10 @@
 
 ;; If signed zeros are ignored, -(a * b + c) = -a * b - c.
 (define_insn "*fnmadd<mode>4"
-  [(set (match_operand:GPF 0 "register_operand")
-	(neg:GPF (fma:GPF (match_operand:GPF 1 "register_operand")
-			  (match_operand:GPF 2 "register_operand")
-			  (match_operand:GPF 3 "register_operand"))))]
+  [(set (match_operand:GPF 0 "register_operand" "=w")
+	(neg:GPF (fma:GPF (match_operand:GPF 1 "register_operand" "w")
+			  (match_operand:GPF 2 "register_operand" "w")
+			  (match_operand:GPF 3 "register_operand" "w"))))]
   "!HONOR_SIGNED_ZEROS (<MODE>mode) && TARGET_FLOAT"
   "fnmadd\\t%<s>0, %<s>1, %<s>2, %<s>3"
   [(set_attr "v8type" "fmadd")
diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index a092dfff9..f1fb239e3 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -4975,7 +4975,7 @@ vbslq_f32 (uint32x4_t a, float32x4_t b, float32x4_t c)
 }
 
 __extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
-vbslq_f64 (float64x2_t a, float64x2_t b, float64x2_t c)
+vbslq_f64 (uint64x2_t a, float64x2_t b, float64x2_t c)
 {
   float64x2_t result;
   __asm__ ("bsl %0.16b, %2.16b, %3.16b"
@@ -7887,6 +7887,39 @@ vfmaq_f64 (float64x2_t a, float64x2_t b, float64x2_t c)
      })
 
 __extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
+vfma_n_f32 (float32x2_t a, float32x2_t b, float32_t c)
+{
+  float32x2_t result;
+  __asm__ ("fmla %0.2s, %2.2s, %3.s[0]"
+           : "=w"(result)
+           : "0"(a), "w"(b), "w"(c)
+           : /* No clobbers */);
+  return result;
+}
+
+__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
+vfmaq_n_f32 (float32x4_t a, float32x4_t b, float32_t c)
+{
+  float32x4_t result;
+  __asm__ ("fmla %0.4s, %2.4s, %3.s[0]"
+           : "=w"(result)
+           : "0"(a), "w"(b), "w"(c)
+           : /* No clobbers */);
+  return result;
+}
+
+__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
+vfmaq_n_f64 (float64x2_t a, float64x2_t b, float64_t c)
+{
+  float64x2_t result;
+  __asm__ ("fmla %0.2d, %2.2d, %3.d[0]"
+           : "=w"(result)
+           : "0"(a), "w"(b), "w"(c)
+           : /* No clobbers */);
+  return result;
+}
+
+__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
 vfms_f32 (float32x2_t a, float32x2_t b, float32x2_t c)
 {
   float32x2_t result;
@@ -14084,102 +14117,6 @@ vqdmulhq_n_s32 (int32x4_t a, int32_t b)
   return result;
 }
 
-#define vqmlalh_lane_s16(a, b, c)                                       \
-  __extension__                                                         \
-    ({                                                                  \
-       int16x8_t b_ = (b);                                              \
-       int16_t a_ = (a);                                                \
-       int32_t result;                                                  \
-       __asm__ ("sqdmlal %s0,%h1,%2.h[%3]"                              \
-                : "=w"(result)                                          \
-                : "w"(a_), "w"(b_), "i"(c)                              \
-                : /* No clobbers */);                                   \
-       result;                                                          \
-     })
-
-__extension__ static __inline int32_t __attribute__ ((__always_inline__))
-vqmlalh_s16 (int16_t a, int16_t b)
-{
-  int32_t result;
-  __asm__ ("sqdmlal %s0,%h1,%h2"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-#define vqmlals_lane_s32(a, b, c)                                       \
-  __extension__                                                         \
-    ({                                                                  \
-       int32x4_t b_ = (b);                                              \
-       int32_t a_ = (a);                                                \
-       int64_t result;                                                  \
-       __asm__ ("sqdmlal %d0,%s1,%2.s[%3]"                              \
-                : "=w"(result)                                          \
-                : "w"(a_), "w"(b_), "i"(c)                              \
-                : /* No clobbers */);                                   \
-       result;                                                          \
-     })
-
-__extension__ static __inline int64_t __attribute__ ((__always_inline__))
-vqmlals_s32 (int32_t a, int32_t b)
-{
-  int64_t result;
-  __asm__ ("sqdmlal %d0,%s1,%s2"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-#define vqmlslh_lane_s16(a, b, c)                                       \
-  __extension__                                                         \
-    ({                                                                  \
-       int16x8_t b_ = (b);                                              \
-       int16_t a_ = (a);                                                \
-       int32_t result;                                                  \
-       __asm__ ("sqdmlsl %s0,%h1,%2.h[%3]"                              \
-                : "=w"(result)                                          \
-                : "w"(a_), "w"(b_), "i"(c)                              \
-                : /* No clobbers */);                                   \
-       result;                                                          \
-     })
-
-__extension__ static __inline int32_t __attribute__ ((__always_inline__))
-vqmlslh_s16 (int16_t a, int16_t b)
-{
-  int32_t result;
-  __asm__ ("sqdmlsl %s0,%h1,%h2"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-#define vqmlsls_lane_s32(a, b, c)                                       \
-  __extension__                                                         \
-    ({                                                                  \
-       int32x4_t b_ = (b);                                              \
-       int32_t a_ = (a);                                                \
-       int64_t result;                                                  \
-       __asm__ ("sqdmlsl %d0,%s1,%2.s[%3]"                              \
-                : "=w"(result)                                          \
-                : "w"(a_), "w"(b_), "i"(c)                              \
-                : /* No clobbers */);                                   \
-       result;                                                          \
-     })
-
-__extension__ static __inline int64_t __attribute__ ((__always_inline__))
-vqmlsls_s32 (int32_t a, int32_t b)
-{
-  int64_t result;
-  __asm__ ("sqdmlsl %d0,%s1,%s2"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
 __extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
 vqmovn_high_s16 (int8x8_t a, int16x8_t b)
 {