modify gtFoldExprConst for casting to integer from floating point overflow

1 files changed, 0 insertions, 23 deletions

diff --git a/src/vm/jithelpers.cpp b/src/vm/jithelpers.cpp
index d3620780ed..276c2d6c58 100644
--- a/src/vm/jithelpers.cpp
+++ b/src/vm/jithelpers.cpp
@@ -613,7 +613,6 @@ HCIMPL1_V(UINT64, JIT_Dbl2ULng, double val)
 
     const double two63  = 2147483648.0 * 4294967296.0;
     UINT64 ret;
-#ifdef _TARGET_XARCH_
     if (val < two63) {
         ret = FastDbl2Lng(val);
     }
@@ -621,28 +620,6 @@ HCIMPL1_V(UINT64, JIT_Dbl2ULng, double val)
         // subtract 0x8000000000000000, do the convert then add it back again
         ret = FastDbl2Lng(val - two63) + I64(0x8000000000000000);
     }
-#else
-// In x86/x64, conversion result of negative double to unsigned integer is
-// bit-equivalent unsigned value.
-// But other architecture's compiler convert negative doubles to zero when
-// the target is unsigned.
-    if (!_finite(val)) {
-        ret = UINT64(val);
-    }
-    else {
-        if (val >= 0.0) {
-            if (val < two63) {
-                ret = UINT64(val);
-            }
-            else {
-                ret = FastDbl2Lng(val - two63) + I64(0x8000000000000000);
-            }
-        }
-        else {
-            ret = UINT64(val);
-        }
-    }
-#endif
     return ret;
 }
 HCIMPLEND