More robust conversions.

Author: igankevich

libc-test/test/makedev.rs

@@ -23,10 +23,16 @@ mod t {
     fn compare(major: c_uint, minor: c_uint) {
         let dev = unsafe { makedev_ffi(major, minor) };
         assert_eq!(libc::makedev(major, minor), dev);
-        let major = unsafe { major_ffi(dev) };
-        let minor = unsafe { minor_ffi(dev) };
-        assert_eq!(libc::major(dev), major.try_into().expect("Conversion should succeed"));
-        assert_eq!(libc::minor(dev), minor.try_into().expect("Conversion should succeed"));
+        let expected_major = unsafe { major_ffi(dev) };
+        let expected_minor = unsafe { minor_ffi(dev) };
+        let expected_major_i64: i64 = expected_major.try_into().unwrap();
+        let expected_minor_i64: i64 = expected_minor.try_into().unwrap();
+        let actual_major = libc::major(dev);
+        let actual_minor = libc::minor(dev);
+        let actual_major_i64: i64 = actual_major.try_into().unwrap();
+        let actual_minor_i64: i64 = actual_minor.try_into().unwrap();
+        assert_eq!(actual_major_i64, expected_major_i64);
+        assert_eq!(actual_minor_i64, expected_minor_i64);
     }
 
     // Every OS should be able to handle 8 bit major and minor numbers