Fix HLSL ByteAddressBuffer Load* parameter integer type (#7117)

* Fix HLSL ByteAddressBuffer Load* parameter integer type

* Fix tests

* Fix load with alignment function signature clash

* Fix LoadAligned tests

7 files changed, 85 insertions, 78 deletions

diff --git a/source/slang/hlsl.meta.slang b/source/slang/hlsl.meta.slang
index 6fb3af7fd..99f83776e 100644
--- a/source/slang/hlsl.meta.slang
+++ b/source/slang/hlsl.meta.slang
@@ -180,7 +180,7 @@ struct ByteAddressBuffer
         {
         case hlsl: __intrinsic_asm ".Load";
         default:
-            return __byteAddressBufferLoad<uint>(this, location, 0);
+            return __byteAddressBufferLoad<uint>(this, uint(location), 0);
         }
     }
 
@@ -213,7 +213,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint2 Load2(int location)
+    uint2 Load2(uint location)
     {
         __target_switch
         {
@@ -226,7 +226,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint2 Load2(int location, int alignment)
+    uint2 Load2Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -239,7 +239,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(hlsl, byteaddressbuffer)]
-    uint2 Load2(int location, out uint status)
+    uint2 Load2(uint location, out uint status)
     {
         __target_switch
         {
@@ -255,7 +255,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint2 Load2Aligned(int location)
+    uint2 Load2Aligned(uint location)
     {
         __target_switch
         {
@@ -282,7 +282,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint3 Load3(int location)
+    uint3 Load3(uint location)
     {
         __target_switch
         {
@@ -295,7 +295,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint3 Load3(int location, int alignment)
+    uint3 Load3Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -308,7 +308,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(hlsl, byteaddressbuffer)]
-    uint3 Load3(int location, out uint status)
+    uint3 Load3(uint location, out uint status)
     {
         __target_switch
         {
@@ -323,7 +323,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint3 Load3Aligned(int location)
+    uint3 Load3Aligned(uint location)
     {
         __target_switch
         {
@@ -349,7 +349,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint4 Load4(int location)
+    uint4 Load4(uint location)
     {
         __target_switch
         {
@@ -362,7 +362,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint4 Load4(int location, int alignment)
+    uint4 Load4Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -375,7 +375,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(hlsl, byteaddressbuffer)]
-    uint4 Load4(int location, out uint status)
+    uint4 Load4(uint location, out uint status)
     {
         __target_switch
         {
@@ -390,7 +390,7 @@ struct ByteAddressBuffer
     [__readNone]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer)]
-    uint4 Load4Aligned(int location)
+    uint4 Load4Aligned(uint location)
     {
         __target_switch
         {
@@ -402,14 +402,14 @@ struct ByteAddressBuffer
 
     [__readNone]
     [ForceInline]
-    T Load<T>(int location)
+    T Load<T>(uint location)
     {
         return __byteAddressBufferLoad<T>(this, location, 0);
     }
 
     [__readNone]
     [ForceInline]
-    T Load<T>(int location, int alignment)
+    T LoadAligned<T>(uint location, uint alignment)
     {
         return __byteAddressBufferLoad<T>(this, location, alignment);
     }
@@ -421,7 +421,7 @@ struct ByteAddressBuffer
     ///Currently, this function only supports when `T` is scalar, vector or matrix type.
     [__readNone]
     [ForceInline]
-    T LoadAligned<T>(int location)
+    T LoadAligned<T>(uint location)
     {
         return __byteAddressBufferLoad<T>(this, location, __naturalStrideOf<T>());
     }
@@ -4785,23 +4785,23 @@ uint64_t __asuint64(uint2 i)
 
 __intrinsic_op($(kIROp_ByteAddressBufferLoad))
 [require(cpp_cuda_glsl_hlsl_metal_spirv_wgsl, byteaddressbuffer)]
-T __byteAddressBufferLoad<T>(ByteAddressBuffer buffer, int offset, int alignment);
+T __byteAddressBufferLoad<T>(ByteAddressBuffer buffer, uint offset, uint alignment);
 
 __intrinsic_op($(kIROp_ByteAddressBufferLoad))
 [require(cpp_cuda_glsl_hlsl_metal_spirv_wgsl, byteaddressbuffer_rw)]
-T __byteAddressBufferLoad<T>(RWByteAddressBuffer buffer, int offset, int alignment);
+T __byteAddressBufferLoad<T>(RWByteAddressBuffer buffer, uint offset, uint alignment);
 
 __intrinsic_op($(kIROp_ByteAddressBufferLoad))
 [require(cpp_cuda_glsl_hlsl_metal_spirv_wgsl, byteaddressbuffer_rw)]
-T __byteAddressBufferLoad<T>(RasterizerOrderedByteAddressBuffer buffer, int offset, int alignment);
+T __byteAddressBufferLoad<T>(RasterizerOrderedByteAddressBuffer buffer, uint offset, uint alignment);
 
 __intrinsic_op($(kIROp_ByteAddressBufferStore))
 [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-void __byteAddressBufferStore<T>(RWByteAddressBuffer buffer, int offset, int alignment, T value);
+void __byteAddressBufferStore<T>(RWByteAddressBuffer buffer, uint offset, uint alignment, T value);
 
 __intrinsic_op($(kIROp_ByteAddressBufferStore))
 [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-void __byteAddressBufferStore<T>(RasterizerOrderedByteAddressBuffer buffer, int offset, int alignment, T value);
+void __byteAddressBufferStore<T>(RasterizerOrderedByteAddressBuffer buffer, uint offset, uint alignment, T value);
 
 __intrinsic_op($(kIROp_GetUntypedBufferPtr))
 [require(spirv, byteaddressbuffer)]
@@ -5053,7 +5053,7 @@ struct $(item.name)
         {
         case hlsl: __intrinsic_asm ".Load";
         default:
-            return __byteAddressBufferLoad<uint>(this, location, 0);
+            return __byteAddressBufferLoad<uint>(this, uint(location), 0);
         }
     }
 
@@ -5086,7 +5086,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint2 Load2(int location)
+    uint2 Load2(uint location)
     {
         __target_switch
         {
@@ -5099,7 +5099,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint2 Load2(int location, int alignment)
+    uint2 Load2Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -5116,7 +5116,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint2 Load2Aligned(int location)
+    uint2 Load2Aligned(uint location)
     {
         __target_switch
         {
@@ -5129,7 +5129,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(hlsl, byteaddressbuffer_rw)]
-    uint2 Load2(int location, out uint status)
+    uint2 Load2(uint location, out uint status)
     {
         __target_switch
         {
@@ -5154,7 +5154,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint3 Load3(int location)
+    uint3 Load3(uint location)
     {
         __target_switch
         {
@@ -5167,7 +5167,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint3 Load3(int location, int alignment)
+    uint3 Load3Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -5184,7 +5184,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint3 Load3Aligned(int location)
+    uint3 Load3Aligned(uint location)
     {
         __target_switch
         {
@@ -5197,7 +5197,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(hlsl, byteaddressbuffer_rw)]
-    uint3 Load3(int location, out uint status)
+    uint3 Load3(uint location, out uint status)
     {
         __target_switch
         {
@@ -5221,7 +5221,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint4 Load4(int location)
+    uint4 Load4(uint location)
     {
         __target_switch
         {
@@ -5234,7 +5234,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint4 Load4(int location, int alignment)
+    uint4 Load4Aligned(uint location, uint alignment)
     {
         __target_switch
         {
@@ -5251,7 +5251,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    uint4 Load4Aligned(int location)
+    uint4 Load4Aligned(uint location)
     {
         __target_switch
         {
@@ -5264,7 +5264,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(hlsl, byteaddressbuffer_rw)]
-    uint4 Load4(int location, out uint status)
+    uint4 Load4(uint location, out uint status)
     {
         __target_switch
         {
@@ -5275,7 +5275,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    T Load<T>(int location)
+    T Load<T>(uint location)
     {
         return __byteAddressBufferLoad<T>(this, location, 0);
     }
@@ -5283,7 +5283,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    T Load<T>(int location, int alignment)
+    T LoadAligned<T>(uint location, uint alignment)
     {
         return __byteAddressBufferLoad<T>(this, location, alignment);
     }
@@ -5296,7 +5296,7 @@ struct $(item.name)
     [__NoSideEffect]
     [ForceInline]
     [require(cpp_cuda_glsl_hlsl_metal_spirv, byteaddressbuffer_rw)]
-    T LoadAligned<T>(int location)
+    T LoadAligned<T>(uint location)
     {
         return __byteAddressBufferLoad<T>(this, location, __naturalStrideOf<T>());
     }
diff --git a/tests/compute/byte-address-buffer-align-error.slang b/tests/compute/byte-address-buffer-align-error.slang
index 34300d7c3..11efa561c 100644
--- a/tests/compute/byte-address-buffer-align-error.slang
+++ b/tests/compute/byte-address-buffer-align-error.slang
@@ -17,7 +17,7 @@ void computeMain(uint3 threadId : SV_DispatchThreadID)
 {
     // CHECK: error 41300: invalid alignment `{{.*}}` specified for the byte address buffer resource with the element size of `{{.*}}`
     // CHECK: error 41300: invalid alignment `{{.*}}` specified for the byte address buffer resource with the element size of `{{.*}}`
-    buffer.Store<Block>(0, buffer.Load<Block>(1, 5));
+    buffer.Store<Block>(0, buffer.LoadAligned<Block>(1, 5));
     buffer.Store<Block>(1, buffer.Load<Block>(0), 3);
 
 }
diff --git a/tests/compute/byte-address-buffer-aligned.slang b/tests/compute/byte-address-buffer-aligned.slang
index 0e848b75c..a65444ec2 100644
--- a/tests/compute/byte-address-buffer-aligned.slang
+++ b/tests/compute/byte-address-buffer-aligned.slang
@@ -33,26 +33,26 @@ void computeMain(uint3 threadId : SV_DispatchThreadID)
     // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = {{.*}};
     // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = {{.*}};
 
-    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(int(32));
-    // CHECK2: buffer0_0.Store(int(32),{{.*}});
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(8));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(12));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(16));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(20));
-    // CHECK2: buffer0_0.Store(int(32),float4({{.*}}, {{.*}}, {{.*}}, {{.*}}));
-    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(int(32));
-    // CHECK2: buffer0_0.Store(int(8),{{.*}}[int(0)]);
-    // CHECK2: buffer0_0.Store(int(12),{{.*}}[int(1)]);
-    // CHECK2: buffer0_0.Store(int(16),{{.*}}[int(2)]);
-    // CHECK2: buffer0_0.Store(int(20),{{.*}}[int(3)]);
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(8));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(12));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(16));
-    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(20));
-    // CHECK2: buffer0_0.Store(int(8),{{.*}});
-    // CHECK2: buffer0_0.Store(int(12),{{.*}});
-    // CHECK2: buffer0_0.Store(int(16),{{.*}});
-    // CHECK2: buffer0_0.Store(int(20),{{.*}});
+    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(32U);
+    // CHECK2: buffer0_0.Store(32U,{{.*}});
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(8U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(12U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(16U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(20U);
+    // CHECK2: buffer0_0.Store(32U,float4({{.*}}, {{.*}}, {{.*}}, {{.*}}));
+    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(32U);
+    // CHECK2: buffer0_0.Store(8U,{{.*}}[int(0)]);
+    // CHECK2: buffer0_0.Store(12U,{{.*}}[int(1)]);
+    // CHECK2: buffer0_0.Store(16U,{{.*}}[int(2)]);
+    // CHECK2: buffer0_0.Store(20U,{{.*}}[int(3)]);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(8U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(12U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(16U);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(20U);
+    // CHECK2: buffer0_0.Store(8U,{{.*}});
+    // CHECK2: buffer0_0.Store(12U,{{.*}});
+    // CHECK2: buffer0_0.Store(16U,{{.*}});
+    // CHECK2: buffer0_0.Store(20U,{{.*}});
 
     // CHECK3-DAG: %[[v4f:[a-zA-Z0-9_]+]] = OpTypeVector %float 4
     // CHECK3-DAG: %[[SBv4f:[a-zA-Z0-9_]+]] = OpTypePointer StorageBuffer %[[v4f]]
diff --git a/tests/compute/byte-address-buffer-array.slang b/tests/compute/byte-address-buffer-array.slang
index 429d24072..0231919e5 100644
--- a/tests/compute/byte-address-buffer-array.slang
+++ b/tests/compute/byte-address-buffer-array.slang
@@ -19,21 +19,21 @@ void computeMain(uint3 threadId : SV_DispatchThreadID)
 {
     // CHECK-NOT: warning
 
-    // CHECK2: float4 {{.*}}[int(2)] = (buffer_0).Load<float4 [int(2)] >(int(0));
-    // CHECK2: buffer_0.Store(int(0),{{.*}});
-    // CHECK2: float {{.*}} = (buffer_0).Load<float >(int(4));
-    // CHECK2: float {{.*}} = (buffer_0).Load<float >(int(8));
-    // CHECK2: float {{.*}} = (buffer_0).Load<float >(int(12));
-    // CHECK2: float {{.*}} = (buffer_0).Load<float >(int(16));
+    // CHECK2: float4 {{.*}}[int(2)] = (buffer_0).Load<float4 [int(2)] >(0U);
+    // CHECK2: buffer_0.Store(0U,{{.*}});
+    // CHECK2: float {{.*}} = (buffer_0).Load<float >(4U);
+    // CHECK2: float {{.*}} = (buffer_0).Load<float >(8U);
+    // CHECK2: float {{.*}} = (buffer_0).Load<float >(12U);
+    // CHECK2: float {{.*}} = (buffer_0).Load<float >(16U);
     // CHECK2: float4 {{.*}} = float4({{.*}}, {{.*}}, {{.*}}, {{.*}});
-    // CHECK2: float4 {{.*}} = (buffer_0).Load<float4 >(int(20));
-    // CHECK2: buffer_0.Store(int(16),{{.*}});
-    // CHECK2: buffer_0.Store(int(32),{{.*}});
+    // CHECK2: float4 {{.*}} = (buffer_0).Load<float4 >(20U);
+    // CHECK2: buffer_0.Store(16U,{{.*}});
+    // CHECK2: buffer_0.Store(32U,{{.*}});
 
     // CHECK3: OpAccessChain
     // CHECK3: OpLoad %v4float
     // CHECK3: OpStore
     buffer.Store(0, buffer.LoadAligned<Block>(0));
-    buffer.StoreAligned(16, buffer.Load<Block>(4, 16));
+    buffer.StoreAligned(16, buffer.LoadAligned<Block>(4, 16));
 }
 
diff --git a/tests/compute/byte-address-buffer.slang b/tests/compute/byte-address-buffer.slang
index 65356ec22..3bf811945 100644
--- a/tests/compute/byte-address-buffer.slang
+++ b/tests/compute/byte-address-buffer.slang
@@ -5,6 +5,8 @@
 //TEST(compute, vulkan):COMPARE_COMPUTE_EX:-vk -compute -shaderobj
 //TEST(compute):COMPARE_COMPUTE_EX:-d3d12 -compute -shaderobj
 
+//TEST:SIMPLE(filecheck=SPV):-target spirv -entry computeMain -stage compute -emit-spirv-directly
+
 // Confirm cross-compilation of `(RW)ByteAddressBuffer`
 //
 // TODO: I'm only using `RWByteAddressBuffer` for now because I don't
@@ -16,26 +18,31 @@ RWByteAddressBuffer inputBuffer;
 //TEST_INPUT:ubuffer(data=[0 0 0 0]):out,name=outputBuffer
 RWByteAddressBuffer outputBuffer;
 
-void test(int val)
+// Make sure arithmetic ops are properly performed on unsigned types.
+// SPV-NOT: OpSDiv
+// SPV: OpUDiv
+
+void test(uint val)
 {
 	uint tmp = val;
 
-	tmp = inputBuffer.Load(int(tmp * 4));
+	tmp = inputBuffer.Load(uint(tmp * 4));
 
-	uint2 pair = inputBuffer.Load2(int(tmp * 4));
+	uint2 pair = inputBuffer.Load2(uint(tmp * 4));
 	tmp = (pair.x + pair.y) & 0xF;
 
-	uint4 quad = inputBuffer.Load4(int(tmp * 4));
+	uint4 quad = inputBuffer.Load4(uint(tmp * 4));
 	tmp = (quad.x + quad.y + quad.z + quad.w) & 0xF;
 
 	outputBuffer.Store(val*4, tmp);
 }
 
 [numthreads(4, 1, 1)]
+[shader("compute")]
 void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
 {
     uint tid = dispatchThreadID.x;
 
-    int val = int(tid);
+    uint val = uint(tid);
     test(val);
 }
diff --git a/tests/metal/byte-address-buffer.slang b/tests/metal/byte-address-buffer.slang
index cf704f170..53c0e0ac5 100644
--- a/tests/metal/byte-address-buffer.slang
+++ b/tests/metal/byte-address-buffer.slang
@@ -19,12 +19,12 @@ struct TestStruct
 [numthreads(1,1,1)]
 void main_kernel(uint3 tid: SV_DispatchThreadID)
 {
-    // CHECK: uint [[WORD0:[a-zA-Z0-9_]+]] = as_type<uint>({{.*}}[(int(0))>>2]);
+    // CHECK: uint [[WORD0:[a-zA-Z0-9_]+]] = as_type<uint>({{.*}}[(0U)>>2]);
     // CHECK: uint8_t [[A:[a-zA-Z0-9_]+]] = uint8_t(([[WORD0]] >> 0U) & 255U);
-    // CHECK: uint [[WORD1:[a-zA-Z0-9_]+]] = as_type<uint>({{.*}}[(int(0))>>2]);
+    // CHECK: uint [[WORD1:[a-zA-Z0-9_]+]] = as_type<uint>({{.*}}[(0U)>>2]);
     // CHECK: half [[H:[a-zA-Z0-9_]+]] = as_type<half>(ushort(([[WORD1]] >> 16U) & 65535U));
 
-    // CHECK: {{.*}}[(int(128))>>2] = as_type<uint32_t>(({{.*}} & 4294967040U) | (uint([[A]]) << 0U));
-    // CHECK: {{.*}}[(int(128))>>2] = as_type<uint32_t>(({{.*}} & 65535U) | (uint(as_type<ushort>([[H]])) << 16U));
+    // CHECK: {{.*}}[(128U)>>2] = as_type<uint32_t>(({{.*}} & 4294967040U) | (uint([[A]]) << 0U));
+    // CHECK: {{.*}}[(128U)>>2] = as_type<uint32_t>(({{.*}} & 65535U) | (uint(as_type<ushort>([[H]])) << 16U));
     buffer.Store(128, buffer.Load<TestStruct>(0));
 }
diff --git a/tests/metal/system-val-conversion.slang b/tests/metal/system-val-conversion.slang
index 5a208086d..dd8e103ae 100644
--- a/tests/metal/system-val-conversion.slang
+++ b/tests/metal/system-val-conversion.slang
@@ -20,7 +20,7 @@ struct TestStruct
 }
 
 // CHECK: void main_kernel(uint3 tid{{.*}}
-// CHECK: int tid{{.*}} = int(tid{{.*}}.x);
+// CHECK: uint(int(tid{{.*}}.x))
 
 [numthreads(1,1,1)]
 void main_kernel(int tid: SV_DispatchThreadID)