JIT: Fix tailcall rehoming logic for some interfering operands

src/coreclr/jit/lower.cpp

@@ -3338,10 +3338,12 @@ void Lowering::LowerFastTailCall(GenTreeCall* call)
     GenTree* startNonGCNode = nullptr;
     if (!putargs.Empty())
     {
-        GenTree* firstPutargStk = putargs.Bottom(0);
+        GenTree* firstPutargStk   = putargs.Bottom(0);
+        GenTree* firstPutargStkOp = firstPutargStk->gtGetOp1();
         for (int i = 1; i < putargs.Height(); i++)
         {
-            firstPutargStk = LIR::FirstNode(firstPutargStk, putargs.Bottom(i));
+            firstPutargStk   = LIR::FirstNode(firstPutargStk, putargs.Bottom(i));
+            firstPutargStkOp = LIR::FirstNode(firstPutargStkOp, putargs.Bottom(i)->gtGetOp1());
         }
         // Since this is a fast tailcall each PUTARG_STK will place the argument in the
         // _incoming_ arg space area. This will effectively overwrite our already existing
@@ -3389,10 +3391,10 @@ void Lowering::LowerFastTailCall(GenTreeCall* call)
                 GenTree* lookForUsesFrom = put->gtNext;
                 if (overwrittenStart != argStart)
                 {
-                    lookForUsesFrom = firstPutargStk;
+                    lookForUsesFrom = firstPutargStkOp;
                 }
 
-                RehomeArgForFastTailCall(callerArgLclNum, firstPutargStk, lookForUsesFrom, call);
+                RehomeArgForFastTailCall(callerArgLclNum, firstPutargStkOp, lookForUsesFrom, call);
                 // The above call can introduce temps and invalidate the pointer.
                 callerArgDsc = comp->lvaGetDesc(callerArgLclNum);
 
@@ -3406,7 +3408,7 @@ void Lowering::LowerFastTailCall(GenTreeCall* call)
                 unsigned int fieldsEnd   = fieldsFirst + callerArgDsc->lvFieldCnt;
                 for (unsigned int j = fieldsFirst; j < fieldsEnd; j++)
                 {
-                    RehomeArgForFastTailCall(j, firstPutargStk, lookForUsesFrom, call);
+                    RehomeArgForFastTailCall(j, firstPutargStkOp, lookForUsesFrom, call);
                 }
             }
         }

src/tests/JIT/Regression/JitBlue/Runtime_116466/Runtime_116466.cs

@@ -0,0 +1,63 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+// Generated by Fuzzlyn v3.2 on 2025-06-08 15:45:13
+// Run on X64 Linux
+// Seed: 5301785820831683815-vectort,vector128,vector256,x86aes,x86avx,x86avx2,x86avx512bw,x86avx512bwvl,x86avx512cd,x86avx512cdvl,x86avx512dq,x86avx512dqvl,x86avx512f,x86avx512fvl,x86avx512fx64,x86bmi1,x86bmi1x64,x86bmi2,x86bmi2x64,x86fma,x86lzcnt,x86lzcntx64,x86pclmulqdq,x86popcnt,x86popcntx64,x86sse,x86ssex64,x86sse2,x86sse2x64,x86sse3,x86sse41,x86sse41x64,x86sse42,x86sse42x64,x86ssse3,x86x86base
+// Reduced from 219.4 KiB to 1.8 KiB in 00:06:09
+// Debug: Outputs 1
+// Release: Outputs 0
+using System;
+using System.Numerics;
+using System.Runtime.CompilerServices;
+using System.Runtime.Intrinsics;
+using Xunit;
+
+public class Runtime_116466
+{
+    static Vector256<uint> s_19;
+    [ThreadStatic]
+    static sbyte s_25;
+    static int s_result;
+
+    [Fact]
+    public static int TestEntryPoint()
+    {
+        s_25 = 123;
+        var vr3 = new S0(100);
+        var vr4 = new S1();
+        M11(vr3, vr4);
+        return s_result;
+    }
+
+    static short M11(S0 arg0, S1 arg7)
+    {
+        s_19 = s_19;
+        var vr2 = Vector128.CreateScalar(0UL).AsVector();
+        return M12(vr2, s_25, arg0);
+    }
+
+    [MethodImpl(MethodImplOptions.NoInlining)]
+    static short M12(Vector<ulong> arg0, sbyte arg2, S0 arg3)
+    {
+        short var1 = arg3.F7;
+        s_result = var1;
+        return arg3.F3;
+    }
+
+    struct S0
+    {
+        public short F3;
+        public Vector<ushort> F6;
+        public short F7;
+        public S0(short f7) : this()
+        {
+            F7 = f7;
+        }
+    }
+
+    struct S1
+    {
+        public S0 F0;
+    }
+}

src/tests/JIT/Regression/JitBlue/Runtime_116466/Runtime_116466.csproj

@@ -0,0 +1,8 @@
+<Project Sdk="Microsoft.NET.Sdk">
+  <PropertyGroup>
+    <Optimize>True</Optimize>
+  </PropertyGroup>
+  <ItemGroup>
+    <Compile Include="$(MSBuildProjectName).cs" />
+  </ItemGroup>
+</Project>