[Beignet] [PATCH] add 64-bit version of "clz"

[Homer Hsing]

this patch passes following piglit test cases:

 piglit/framework/../bin/cl-program-tester generated_tests/cl/builtin/int/builtin-ulong-clz-1.0.generated.cl
 piglit/framework/../bin/cl-program-tester generated_tests/cl/builtin/int/builtin-long-clz-1.0.generated.cl

Signed-off-by: Homer Hsing <homer.xing at intel.com>
---
 backend/src/ocl_stdlib.tmpl.h | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/backend/src/ocl_stdlib.tmpl.h b/backend/src/ocl_stdlib.tmpl.h
index 0a6a937..f209b52 100644
--- a/backend/src/ocl_stdlib.tmpl.h
+++ b/backend/src/ocl_stdlib.tmpl.h
@@ -313,11 +313,27 @@ INLINE_OVERLOADABLE uint clz(uint x) {
 }
 
 INLINE_OVERLOADABLE long clz(long x) {
-  return 0;
+  if (x < 0)
+    return 0;
+  if (x == 0)
+    return 64;
+  union { int i[2]; long x; } u;
+  u.x = x;
+  uint v = clz(u.i[1]);
+  if(v == 32)
+    v += clz(u.i[0]);
+  return v;
 }
 
 INLINE_OVERLOADABLE ulong clz(ulong x) {
-  return 0;
+  if (x == 0)
+    return 64;
+  union { uint i[2]; ulong x; } u;
+  u.x = x;
+  uint v = clz(u.i[1]);
+  if(v == 32)
+    v += clz(u.i[0]);
+  return v;
 }
 
 OVERLOADABLE int __gen_ocl_mul_hi(int x, int y);
-- 
1.8.1.2