Repository navigation
JIT: (bug) a branch on the bitwise complement of a comparison makes a later identical comparison fold to the wrong value #133557
Description
Repro
using System;
using System.Runtime.CompilerServices;
class Program
{
[MethodImpl(MethodImplOptions.NoInlining | MethodImplOptions.AggressiveOptimization)]
static int Test(int a, int b)
{
int t = (a != b) ? 1 : 0;
int u = ~t; // -1 or -2, so u is never zero
if (u != 0)
{
if (a != b)
return 1;
}
return 0;
}
static void Main() => Console.WriteLine(Test(3, 5));
}Expected
1
Actual
0
The outer condition ~t != 0 is a tautology that carries no information about a != b, but the
inner comparison is folded from it anyway.
Platform
Windows x64, local Checked build of dotnet/runtime main (ef97eb5). Not target-specific.
MinOpts and release .NET 11.0.100-rc.1.26420.103 both produce the correct 1.
Activity
dotnet-policy-service commentedon Sep 10, 2026
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.
MihuBot commentedon Sep 10, 2026
I'm a bot. Here is a possible related and/or duplicate issue (I may be wrong):
~x op kwithout inverting the comparison #129076AndyAyersMS commentedon Sep 10, 2026
Bad RBO inferences from GT_NOT. We fixed them in VN, but didn't fix them in RBO. Grr.
JIT: stop inferring through bitwise complement (#133700)
EgorBo commentedon Sep 18, 2026
Affected SDK versions: