fixed uint underflow in floorf for negative exponents
This commit is contained in:
+1
-1
@@ -12,7 +12,7 @@ pub fn floorf(x: f32) -> f32 {
|
||||
}
|
||||
}
|
||||
let mut ui = x.to_bits();
|
||||
let e = (((ui >> 23) & 0xff) - 0x7f) as i32;
|
||||
let e = (((ui >> 23) as i32) & 0xff) - 0x7f;
|
||||
|
||||
if e >= 23 {
|
||||
return x;
|
||||
|
||||
Reference in New Issue
Block a user