• d-bugmail puremagic.com (28/28) Nov 03 2020 https://issues.dlang.org/show_bug.cgi?id=21363

d-bugmail puremagic.com writes:

https://issues.dlang.org/show_bug.cgi?id=21363

          Issue ID: 21363
           Summary: [REG2.094] Implementation of core.bitop.ror(x,0) is
                    using UB
           Product: D
           Version: D2
          Hardware: All
                OS: All
            Status: NEW
          Severity: regression
          Priority: P1
         Component: druntime
          Assignee: nobody puremagic.com
          Reporter: johanengelen weka.io

This is a long-standing bug (in practice it doesn't result in UB), and now
comes out due to better CTFE checking of UB.

Testcase:
```
import core.bitop;
enum a = ror(uint(1), 0); // rotate by 0, should be allowed.
```

The test case no longer compiles with 2.094.0:
core/bitop.d(961): Error: shift by 32 is outside the range 0..31
rortest.d(1):        called from here: ror(1u, 0u)


The problem is that the implementation of `ror` is indeed shifting by a too
large amount.
Note that `rol` has the same problem.

--