[AGC] Correct VReadlaneB32 decode and emission (#237)

VReadlaneB32 (VOP3 0x360) had two bugs:

1. Decode: VOP3 decode always set destinations = Vector(word & 0xFF).
   For VReadlaneB32, bits 0-7 are unused — the scalar destination is
   in bits 8-14. Now decodes as Scalar((word >> 8) & 0x7F).

2. Emission: VReadlaneB32 was in the VMovB32 fall-through group,
   just returning GetRawSource(instruction, 0) — reading the current
   lane's value. By ISA, sdst = vsrc0[lane(src1)], which requires
   reading a different lane's value. Now uses SPIR-V
   GroupNonUniformBroadcast(scope=Subgroup, value=src0, lane=src1)
   when subgroup operations are available, with a fallback to the
   current-lane simplification when not.

3. Routing: TryEmitVectorAlu called TryGetVectorDestination first,
   which checks for VectorRegister kind. With the scalar destination
   fix, VReadlaneB32 now routes to a new TryEmitReadlane handler
   before the vector destination check.

4. Subgroup capability: UsesSubgroupShuffle now includes VReadlaneB32
   so GroupNonUniform capability is enabled when needed.

Verified: dotnet build 0 errors/0 warnings, 26/26 tests pass,
ShaderDump all programs behaved as expected.

src/SharpEmu.ShaderCompiler.Vulkan/Gen5SpirvTranslator.Alu.cs

@@ -24,6 +24,11 @@ public static partial class Gen5SpirvTranslator
                 return TryEmitVectorCompare(instruction, out error);
             }
 
+            if (instruction.Opcode == "VReadlaneB32")
+            {
+                return TryEmitReadlane(instruction, out error);
+            }
+
             if (!TryGetVectorDestination(instruction, out var destination))
             {
                 error = "missing vector destination";
@@ -34,7 +39,6 @@ public static partial class Gen5SpirvTranslator
             switch (instruction.Opcode)
             {
                 case "VMovB32":
-                case "VReadlaneB32":
                 case "VReadfirstlaneB32":
                     result = GetRawSource(instruction, 0);
                     break;
@@ -2365,6 +2369,42 @@ public static partial class Gen5SpirvTranslator
             StoreWaveMask(106, carry);
         }
 
+        private bool TryEmitReadlane(
+            Gen5ShaderInstruction instruction,
+            out string error)
+        {
+            error = string.Empty;
+            if (instruction.Destinations.Count == 0 ||
+                instruction.Destinations[0].Kind != Gen5OperandKind.ScalarRegister)
+            {
+                error = "VReadlaneB32 expects scalar destination";
+                return false;
+            }
+
+            var destination = instruction.Destinations[0].Value;
+            var src0 = GetRawSource(instruction, 0);
+
+            if (_subgroupInvocationIdInput != 0)
+            {
+                // sdst = vsrc0[lane(src1)] — broadcast from the specified lane.
+                var laneSelect = GetRawSource(instruction, 1);
+                var broadcast = _module.AddInstruction(
+                    SpirvOp.GroupNonUniformBroadcast,
+                    _uintType,
+                    UInt(3),  // Subgroup scope
+                    src0,
+                    laneSelect);
+                StoreS(destination, broadcast);
+            }
+            else
+            {
+                // Fallback: no subgroup ops, read current lane's value.
+                StoreS(destination, src0);
+            }
+
+            return true;
+        }
+
         private uint EmitPermlane16(
             Gen5ShaderInstruction instruction,
             bool exchangeRows)

src/SharpEmu.ShaderCompiler.Vulkan/Gen5SpirvTranslator.cs

@@ -2507,7 +2507,7 @@ public static partial class Gen5SpirvTranslator
 
         private bool UsesSubgroupShuffle() =>
             _state.Program.Instructions.Any(instruction =>
-                instruction.Opcode is "VPermlane16B32" or "VPermlanex16B32");
+                instruction.Opcode is "VPermlane16B32" or "VPermlanex16B32" or "VReadlaneB32");
 
         private bool UsesWaveControl() =>
             _state.Program.Instructions.Any(instruction =>

src/SharpEmu.ShaderCompiler/Gen5ShaderTranslator.cs

@@ -1467,6 +1467,12 @@ public static class Gen5ShaderTranslator
                     Gen5Operand.Source((extra >> 18) & 0x1FF, literal),
                 ];
                 destinations = [Gen5Operand.Vector(word & 0xFF)];
+                if (opcode == "VReadlaneB32")
+                {
+                    // VReadlaneB32 writes to scalar destination (bits 8-14), not vector.
+                    // Bits 0-7 are unused for this opcode.
+                    destinations = [Gen5Operand.Scalar((word >> 8) & 0x7F)];
+                }
                 var isVop3B = IsVop3BOpcode((word >> 16) & 0x3FF);
                 control = new Gen5Vop3Control(
                     isVop3B ? 0 : (word >> 8) & 0x7,