target-tilegx: Implement v*shl, v*shru, and v*shrs instructions

v2sh* are implemented with helper functions; v4sh* are implmeneted
with inline code.

Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Message-Id: <1442872055-2836-1-git-send-email-gang.chen.5i5j@gmail.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>

1 files changed, 31 insertions, 0 deletions

diff --git a/target-tilegx/simd_helper.c b/target-tilegx/simd_helper.c
index f573f9b51a..1c59a92216 100644
--- a/target-tilegx/simd_helper.c
+++ b/target-tilegx/simd_helper.c
@@ -25,6 +25,7 @@
 
 /* Broadcast a value to all elements of a vector.  */
 #define V1(X)      (((X) & 0xff) * 0x0101010101010101ull)
+#define V2(X)      (((X) & 0xffff) * 0x0001000100010001ull)
 
 
 uint64_t helper_v1shl(uint64_t a, uint64_t b)
@@ -36,6 +37,15 @@ uint64_t helper_v1shl(uint64_t a, uint64_t b)
     return (a & m) << b;
 }
 
+uint64_t helper_v2shl(uint64_t a, uint64_t b)
+{
+    uint64_t m;
+
+    b &= 15;
+    m = V2(0xffff >> b);
+    return (a & m) << b;
+}
+
 uint64_t helper_v1shru(uint64_t a, uint64_t b)
 {
     uint64_t m;
@@ -45,6 +55,15 @@ uint64_t helper_v1shru(uint64_t a, uint64_t b)
     return (a & m) >> b;
 }
 
+uint64_t helper_v2shru(uint64_t a, uint64_t b)
+{
+    uint64_t m;
+
+    b &= 15;
+    m = V2(0xffff << b);
+    return (a & m) >> b;
+}
+
 uint64_t helper_v1shrs(uint64_t a, uint64_t b)
 {
     uint64_t r = 0;
@@ -56,3 +75,15 @@ uint64_t helper_v1shrs(uint64_t a, uint64_t b)
     }
     return r;
 }
+
+uint64_t helper_v2shrs(uint64_t a, uint64_t b)
+{
+    uint64_t r = 0;
+    int i;
+
+    b &= 15;
+    for (i = 0; i < 64; i += 16) {
+        r = deposit64(r, i, 16, sextract64(a, i + b, 16 - b));
+    }
+    return r;
+}