Update llvm-project to llvm/llvm-project@01d233ff403823389f848 (#3011)

Co-authored-by: Megan Hampton <[email protected]>

docs/BuildOnLinuxOSX.md

@@ -15,7 +15,7 @@ Firstly, install MLIR (as a part of LLVM-Project):
 ``` bash
 git clone -n https://github.com/llvm/llvm-project.git
 # Check out a specific branch that is known to work with ONNX-MLIR.
-cd llvm-project && git checkout 00128a20eec27246719d73ba427bf821883b00b4 && cd ..
+cd llvm-project && git checkout 01d233ff403823389f8480897e41aea84ecbb3d3 && cd ..
 ```
 
 [same-as-file]: <> (utils/build-mlir.sh)

docs/BuildOnWindows.md

@@ -52,7 +52,7 @@ Install MLIR (as a part of LLVM-Project):
 ```shell
 git clone -n https://github.com/llvm/llvm-project.git
 # Check out a specific branch that is known to work with ONNX-MLIR.
-cd llvm-project && git checkout 00128a20eec27246719d73ba427bf821883b00b4 && cd ..
+cd llvm-project && git checkout 01d233ff403823389f8480897e41aea84ecbb3d3 && cd ..
 ```
 
 [same-as-file]: <> (utils/build-mlir.cmd)

docs/Dialects/krnl.md

@@ -329,6 +329,7 @@ _Indicate ONNX entry point_
 
 The "krnl.entry_point" function indicates the main entry
                            point of ONNX model.
+
 ### `krnl.erf` (KrnlErfOp)
 
 _Krnl erf scalar operation_
@@ -453,7 +454,7 @@ in the `value` dense element attribute.
 
 Traits: `AlwaysSpeculatableImplTrait`, `MemRefsNormalizable`
 
-Interfaces: `ConditionallySpeculatable`, `KrnlGlobalOpInterface`, `NoMemoryEffect (MemoryEffectOpInterface)`
+Interfaces: `ConditionallySpeculatable`, `NoMemoryEffect (MemoryEffectOpInterface)`
 
 Effects: `MemoryEffects::Effect{}`
 

docs/Dialects/zhigh.md

@@ -793,8 +793,6 @@ Effects: `MemoryEffects::Effect{}`
 _ZHigh Stickified Constant operation_
 
 This operator produces a constant tensor to store stickified data.
-`value` attribute has original constant or stickified constant.
-`stickified` attribute indicates the `value` is already stickified or not.
 Stickified data is opaque and must be 4K-aligned. One who produces
 the stickified data must make sure its size in bytes consistent with
 the output tensor's size.
@@ -809,7 +807,6 @@ Effects: `MemoryEffects::Effect{}`
 
 <table>
 <tr><th>Attribute</th><th>MLIR Type</th><th>Description</th></tr>
-<tr><td><code>stickified</code></td><td>::mlir::BoolAttr</td><td>bool attribute</td></tr>
 <tr><td><code>value</code></td><td>::mlir::Attribute</td><td>any attribute</td></tr>
 <tr><td><code>alignment</code></td><td>::mlir::IntegerAttr</td><td>64-bit signless integer attribute</td></tr>
 </table>

docs/Dialects/zlow.md

@@ -752,34 +752,6 @@ Interfaces: `MemoryEffectOpInterface`
 | `X` | memref of 16-bit float or 32-bit float values
 | `Out` | memref of dlfloat16 type values
 
-### `zlow.stickifiedConstant` (::onnx_mlir::zlow::ZLowStickifiedConstantOp)
-
-_ZLow Stickified Constant operation._
-
-
-Traits: `MemRefsNormalizable`
-
-Interfaces: `KrnlGlobalOpInterface`
-
-#### Attributes:
-
-<table>
-<tr><th>Attribute</th><th>MLIR Type</th><th>Description</th></tr>
-<tr><td><code>shape</code></td><td>::mlir::Attribute</td><td>any attribute</td></tr>
-<tr><td><code>name</code></td><td>::mlir::StringAttr</td><td>string attribute</td></tr>
-<tr><td><code>stickified</code></td><td>::mlir::BoolAttr</td><td>bool attribute</td></tr>
-<tr><td><code>value</code></td><td>::mlir::Attribute</td><td>any attribute</td></tr>
-<tr><td><code>layout</code></td><td>::mlir::StringAttr</td><td>string attribute</td></tr>
-<tr><td><code>offset</code></td><td>::mlir::IntegerAttr</td><td>64-bit signless integer attribute</td></tr>
-<tr><td><code>alignment</code></td><td>::mlir::IntegerAttr</td><td>64-bit signless integer attribute</td></tr>
-</table>
-
-#### Results:
-
-| Result | Description |
-| :----: | ----------- |
-| `output` | memref of dlfloat16 type values
-
 ### `zlow.sub` (::onnx_mlir::zlow::ZLowSubOp)
 
 _ZLow sub operation_

src/Accelerators/NNPA/Transform/ZLow/ZLowStickExpansion.cpp

@@ -153,9 +153,7 @@ class UnstickExpansionPattern : public OpRewritePattern<ZLowUnstickOp> {
     // 64.
     IndexExpr T = LitIE(2);
     DimsExpr reallocTileDims = {T, lit64};
-    Value inputAsTx64 =
-        create.mem.reinterpretCast(input, litZero.getValue(), reallocTileDims);
-
+    Value inputAsTx64 = create.mem.reinterpretCast(input, reallocTileDims);
     // Outer loop (E4, E3, E2, E1 iterates over tiles of 64 elements)
     create.krnl.iterateIE(loopDefs, loopDefs, lbs, ubs,
         [&](const KrnlBuilder &b, ValueRange loopInd) {
@@ -456,8 +454,7 @@ class StickExpansionPattern : public OpRewritePattern<ZLowStickOp> {
     // 64 elements.
     IndexExpr T = LitIE(2);
     DimsExpr reallocTileDims = {T, lit64};
-    Value allocAsTx64 =
-        create.mem.reinterpretCast(alloc, litZero.getValue(), reallocTileDims);
+    Value allocAsTx64 = create.mem.reinterpretCast(alloc, reallocTileDims);
 
     // Outer loop (E1 iterates over tiles of 64 elements).
     create.krnl.iterateIE(loopDefs, loopDefs, lbs, ubs,

src/Conversion/KrnlToAffine/ConvertKrnlToAffine.cpp

@@ -814,20 +814,18 @@ AffineTypeConverter::AffineTypeConverter() {
   addConversion([](Type type) { return type; });
 
   addSourceMaterialization([&](OpBuilder &builder, Type resultType,
-                               ValueRange inputs,
-                               Location loc) -> std::optional<Value> {
+                               ValueRange inputs, Location loc) -> Value {
     if (inputs.size() != 1)
-      return std::nullopt;
+      return Value();
 
     return builder.create<UnrealizedConversionCastOp>(loc, resultType, inputs)
         .getResult(0);
   });
 
   addTargetMaterialization([&](OpBuilder &builder, Type resultType,
-                               ValueRange inputs,
-                               Location loc) -> std::optional<Value> {
+                               ValueRange inputs, Location loc) -> Value {
     if (inputs.size() != 1)
-      return std::nullopt;
+      return Value();
 
     return builder.create<UnrealizedConversionCastOp>(loc, resultType, inputs)
         .getResult(0);

src/Conversion/ONNXToKrnl/ONNXToKrnlCommon.cpp

@@ -547,20 +547,18 @@ KrnlTypeConverter::KrnlTypeConverter() {
   });
 
   addSourceMaterialization([&](OpBuilder &builder, Type resultType,
-                               ValueRange inputs,
-                               Location loc) -> std::optional<Value> {
+                               ValueRange inputs, Location loc) -> Value {
     if (inputs.size() != 1)
-      return std::nullopt;
+      return Value();
 
     return builder.create<UnrealizedConversionCastOp>(loc, resultType, inputs)
         .getResult(0);
   });
 
   addTargetMaterialization([&](OpBuilder &builder, Type resultType,
-                               ValueRange inputs,
-                               Location loc) -> std::optional<Value> {
+                               ValueRange inputs, Location loc) -> Value {
     if (inputs.size() != 1)
-      return std::nullopt;
+      return Value();
 
     return builder.create<UnrealizedConversionCastOp>(loc, resultType, inputs)
         .getResult(0);

src/Dialect/Mlir/DialectBuilder.cpp

@@ -617,7 +617,7 @@ Value MathBuilder::constant(Type type, double val) const {
             b().create<arith::ConstantOp>(loc(), b().getF64FloatAttr(val));
       })
       .Case<IntegerType>([&](IntegerType elementType) {
-        assert(val == (int64_t)val && "value is ambiguous");
+        assert(val == static_cast<int64_t>(val) && "value is ambiguous");
         unsigned width = elementType.getWidth();
 
         if (width == 1)
@@ -628,11 +628,13 @@ Value MathBuilder::constant(Type type, double val) const {
           if (elementType.isUnsignedInteger()) {
             Type signlessTy = b().getIntegerType(width);
             constant = b().create<arith::ConstantOp>(loc(),
-                b().getIntegerAttr(signlessTy, APInt(width, (int64_t)val)));
+                b().getIntegerAttr(signlessTy,
+                    APInt(width, static_cast<int64_t>(val), false, true)));
             constant = castToUnsigned(constant, width);
           } else {
             constant = b().create<arith::ConstantOp>(loc(),
-                b().getIntegerAttr(elementType, APInt(width, (int64_t)val)));
+                b().getIntegerAttr(elementType,
+                    APInt(width, static_cast<int64_t>(val), false, true)));
           }
         }
       })
@@ -695,7 +697,7 @@ TypedAttr MathBuilder::negativeInfAttr(Type type) const {
         default:
           llvm_unreachable("unsupported element type");
         }
-        attr = b().getIntegerAttr(type, APInt(width, value));
+        attr = b().getIntegerAttr(type, APInt(width, value, false, true));
       })
       .Default([](Type) { llvm_unreachable("unsupported element type"); });
   assert(attr != nullptr && "Expecting valid attribute");
@@ -740,7 +742,7 @@ TypedAttr MathBuilder::positiveInfAttr(Type type) const {
         default:
           llvm_unreachable("unsupported element type");
         }
-        attr = b().getIntegerAttr(type, APInt(width, value));
+        attr = b().getIntegerAttr(type, APInt(width, value, false, true));
       })
       .Default([](Type) { llvm_unreachable("unsupported element type"); });
   assert(attr != nullptr && "Expecting valid attribute");
@@ -2263,7 +2265,8 @@ Value LLVMBuilder::constant(Type type, int64_t val) const {
           assert(type.isSignless() &&
                  "LLVM::ConstantOp requires a signless type.");
           constant = b().create<LLVM::ConstantOp>(loc(), type,
-              b().getIntegerAttr(type, APInt(width, (int64_t)val)));
+              b().getIntegerAttr(
+                  type, APInt(width, static_cast<int64_t>(val), false, true)));
         }
       })
       .Case<IndexType>([&](Type) {

test/mlir/accelerators/nnpa/driver/matmul-div-in-attention-layer.mlir

@@ -8,13 +8,13 @@ func.func @matmul_div(%arg0: tensor<?x12x?x64xf32>) -> tensor<?x?x?x?xf32> {
   %r = "onnx.Div"(%m, %scalar) : (tensor<?x12x?x?xf32>, tensor<f32>) -> tensor<?x12x?x?xf32>
   "onnx.Return"(%r) : (tensor<?x12x?x?xf32>) -> ()
 
-// CHECK-LABEL:  func.func @matmul_div
+// CHECK-LABEL: func.func @matmul_div
 // CHECK:           memref.alloc
 // CHECK:           memref.alloc
 // CHECK:           [[ALLOC:%.+]] = memref.alloc({{.*}}) {{.*}}: memref<?x?x1x?x32x64xf16>
 // CHECK-DAG:       [[MATMUL_RES:%.+]] = memref.cast [[ALLOC]] : memref<?x?x1x?x32x64xf16> to memref<?x?x1x?x?x?xf16>
 // CHECK:           "zlow.matmul"({{.*}}, {{.*}}, {{.*}}, {{.*}}, [[MATMUL_RES]]) {is_bcast = 0 : si64, is_stacked = -1 : si64} : (memref<?x1x1x?x?x64xf16>, memref<?x?x1x2x32x?xf16>, memref<?x?x1x1x32x?xf16>, memref<4xi64>, memref<?x?x1x?x?x?xf16>) -> ()
-// CHECK-NOT:        "zlow.stick"
-// CHECK-NOT:        "zlow.unstick"
+// CHECK-NOT:       "zlow.stick"
+// CHECK-NOT:       "zlow.unstick"
 // CHECK:           "zlow.div"([[MATMUL_RES]], {{.*}}, {{.*}}, {{.*}}) {layout = "3DS"} : (memref<?x?x1x?x?x?xf16>, memref<?x?x1x?x?x?xf16>, memref<3xi64>, memref<?x?x1x?x?x?xf16>) -> ()
 }

test/mlir/accelerators/nnpa/driver/saturation.mlir

@@ -12,31 +12,31 @@
 func.func @saturation(%arg0 : tensor<10x10xf32>) -> tensor<*xf32> {
   %0 = "onnx.Relu"(%arg0) : (tensor<10x10xf32>) -> tensor<*xf32>
   "func.return"(%0) : (tensor<*xf32>) -> ()
-// ZHIGH_OFF-LABEL:  func @saturation
+// ZHIGH_OFF-LABEL: func @saturation
 // ZHIGH_OFF: "zhigh.Stick"({{.*}}) {layout = "2D"} : {{.*}} 
 
-// ZHIGH_ON-LABEL:  func @saturation
+// ZHIGH_ON-LABEL: func @saturation
 // ZHIGH_ON: "zhigh.Stick"({{.*}}) {layout = "2D", saturation = -1 : si64} : {{.*}} 
 
 
-// ZLOW_OFF-LABEL:  func @saturation
+// ZLOW_OFF-LABEL: func @saturation
 // ZLOW_OFF:   "zlow.stick"({{.*}}, {{.*}}) {layout = "2D"} : {{.*}} 
 
-// ZLOW_ON-LABEL:  func @saturation
+// ZLOW_ON-LABEL: func @saturation
 // ZLOW_ON:   "zlow.stick"({{.*}}, {{.*}}) {layout = "2D", saturation = -1 : si64} : {{.*}} 
 
-// DECOMPOSE_OFF-LABEL:  func @saturation
+// DECOMPOSE_OFF-LABEL: func @saturation
 // DECOMPOSE_OFF: "zhigh.F32ToDLF16"(%arg0) :  {{.*}}
 
-// DECOMPOSE_ON-LABEL:  func @saturation
+// DECOMPOSE_ON-LABEL: func @saturation
 // DECOMPOSE_ON: "zhigh.F32ToDLF16"(%arg0) {saturation = -1 : si64} :  {{.*}}
 
-// COMPILER_STICK_OFF-LABEL:  func @saturation
+// COMPILER_STICK_OFF-LABEL: func @saturation
 // COMPILER_STICK_OFF-NOT: arith.minnumf 
 // COMPILER_STICK_OFF-NOT: arith.maxnumf 
 // COMPILER_STICK_OFF: zlow.relu 
 
-// COMPILER_STICK_ON-LABEL:  func @saturation
+// COMPILER_STICK_ON-LABEL: func @saturation
 // COMPILER_STICK_ON: arith.minnumf 
 // COMPILER_STICK_ON: arith.maxnumf 
 // COMPILER_STICK_ON: zlow.relu 

test/mlir/accelerators/nnpa/module_op_be/compiler-config.mlir

@@ -12,5 +12,5 @@ module {
   "onnx.EntryPoint"() {func = @main_graph} : () -> ()
 }
 // CHECK: {{.*}} opt {{.*}} -o {{.*}}.bc
-// CHECK: {{.*}} llc {{.*}}  {{.*}} {{.*}}.bc
+// CHECK: {{.*}} llc {{.*}} {{.*}} {{.*}}.bc
 // CHECK: {{.*}} {{clang|c|g}}++{{.*}} {{.*}}.o -o {{.*}}.so -shared -fPIC -L{{.*}}/lib -lRuntimeNNPA -lzdnn -lcruntime

test/mlir/accelerators/nnpa/transform/zlow-stick-unstick-expansion.mlir

@@ -28,10 +28,9 @@ func.func @test_stick_expansion_with_sat(%arg0: memref<16x8x128xf32>) -> memref<
 // CHECK-DAG:       [[CST_4_:%.+]] = arith.constant 4 : index
 // CHECK-DAG:       [[VAR_cst_:%.+]] = arith.constant dense<8.57315738E+9> : vector<4xf32>
 // CHECK-DAG:       [[VAR_cst_0_:%.+]] = arith.constant dense<-8.57315738E+9> : vector<4xf32>
-// CHECK-DAG:       [[CST_0_:%.+]] = arith.constant 0 : index
 // CHECK-DAG:       [[RES_:%.+]] = memref.alloc() {{.*}}: memref<16x8x128xf16, #map>
 // CHECK-DAG:       [[LOOP_0_:%.+]]:3 = krnl.define_loops 3
-// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[RES_]] to offset: {{.}}[[CST_0_]]{{.}}, sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
+// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[RES_]] to offset: [0], sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
 // CHECK:           krnl.iterate([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) with ([[LOOP_0_]]#0 -> [[I_0_:%.+]] = 0 to 16, [[LOOP_0_]]#1 -> [[I_1_:%.+]] = 0 to 8, [[LOOP_0_]]#2 -> [[I_2_:%.+]] = 0 to 2){
 // CHECK:             [[VAR_1_:%.+]]:3 = krnl.get_induction_var_value([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) : (!krnl.loop, !krnl.loop, !krnl.loop) -> (index, index, index)
 // CHECK:             [[VAR_2_:%.+]] = affine.apply [[MAP_1_]]([[VAR_1_]]#2)
@@ -124,10 +123,9 @@ func.func @test_stick_expansion_without_sat(%arg0: memref<16x8x128xf32>) -> memr
 // CHECK-DAG:       [[CST_12_:%.+]] = arith.constant 12 : index
 // CHECK-DAG:       [[CST_8_:%.+]] = arith.constant 8 : index
 // CHECK-DAG:       [[CST_4_:%.+]] = arith.constant 4 : index
-// CHECK-DAG:       [[CST_0_:%.+]] = arith.constant 0 : index
 // CHECK-DAG:       [[RES_:%.+]] = memref.alloc() {{.*}}: memref<16x8x128xf16, #map>
 // CHECK-DAG:       [[LOOP_0_:%.+]]:3 = krnl.define_loops 3
-// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[RES_]] to offset: {{.}}[[CST_0_]]{{.}}, sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
+// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[RES_]] to offset: [0], sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
 // CHECK:           krnl.iterate([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) with ([[LOOP_0_]]#0 -> [[I_0_:%.+]] = 0 to 16, [[LOOP_0_]]#1 -> [[I_1_:%.+]] = 0 to 8, [[LOOP_0_]]#2 -> [[I_2_:%.+]] = 0 to 2){
 // CHECK:             [[VAR_1_:%.+]]:3 = krnl.get_induction_var_value([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) : (!krnl.loop, !krnl.loop, !krnl.loop) -> (index, index, index)
 // CHECK:             [[VAR_2_:%.+]] = affine.apply [[MAP_1_]]([[VAR_1_]]#2)
@@ -208,7 +206,7 @@ func.func @test_unstick_expansion(%arg0: memref<16x8x128xf16, #map>) -> memref<1
 // CHECK-DAG:       [[CST_0_:%.+]] = arith.constant 0 : index
 // CHECK-DAG:       [[RES_:%.+]] = memref.alloc() {{.*}}: memref<16x8x128xf32>
 // CHECK-DAG:       [[LOOP_0_:%.+]]:3 = krnl.define_loops 3
-// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[PARAM_0_]] to offset: {{.}}[[CST_0_]]{{.}}, sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
+// CHECK-DAG:       [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[PARAM_0_]] to offset: [0], sizes: [2, 64], strides: [64, 1] : memref<16x8x128xf16, #map> to memref<2x64xf16>
 // CHECK:           krnl.iterate([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) with ([[LOOP_0_]]#0 -> [[I_0_:%.+]] = 0 to 16, [[LOOP_0_]]#1 -> [[I_1_:%.+]] = 0 to 8, [[LOOP_0_]]#2 -> [[I_2_:%.+]] = 0 to 2){
 // CHECK:             [[VAR_1_:%.+]]:3 = krnl.get_induction_var_value([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) : (!krnl.loop, !krnl.loop, !krnl.loop) -> (index, index, index)
 // CHECK:             [[VAR_2_:%.+]] = affine.apply [[MAP_1_]]([[VAR_1_]]#2)
@@ -292,7 +290,7 @@ func.func @test_unstick_expansion_127(%arg0: memref<16x8x127xf16, #map>) -> memr
 // CHECK-DAG:       [[CST_0_:%.+]] = arith.constant 0 : index
 // CHECK-DAG:       [[RES_:%.+]] = memref.alloc() {{.*}}: memref<16x8x127xf32>
 // CHECK-DAG:       [[LOOP_0_:%.+]]:3 = krnl.define_loops 3
-// CHECK:           [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[PARAM_0_]] to offset: {{.}}[[CST_0_]]{{.}}, sizes: [2, 64], strides: [64, 1] : memref<16x8x127xf16, #map> to memref<2x64xf16>
+// CHECK-DAG:       [[VAR_reinterpret_cast_:%.+]] = memref.reinterpret_cast [[PARAM_0_]] to offset: [0], sizes: [2, 64], strides: [64, 1] : memref<16x8x127xf16, #map> to memref<2x64xf16>
 // CHECK:           krnl.iterate([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) with ([[LOOP_0_]]#0 -> [[I_0_:%.+]] = 0 to 16, [[LOOP_0_]]#1 -> [[I_1_:%.+]] = 0 to 8, [[LOOP_0_]]#2 -> [[I_2_:%.+]] = 0 to 2){
 // CHECK:             [[VAR_1_:%.+]]:3 = krnl.get_induction_var_value([[LOOP_0_]]#0, [[LOOP_0_]]#1, [[LOOP_0_]]#2) : (!krnl.loop, !krnl.loop, !krnl.loop) -> (index, index, index)
 // CHECK:             [[VAR_2_:%.+]] = affine.apply [[MAP_1_]]([[VAR_1_]]#2)
@@ -360,5 +358,4 @@ func.func @test_unstick_expansion_127(%arg0: memref<16x8x127xf16, #map>) -> memr
 // CHECK:           }
 // CHECK:           return [[RES_]] : memref<16x8x127xf32>
 // CHECK:         }
-}
-
+}

test/mlir/parallel/krnl_parallel_clause_to_omp.mlir

@@ -32,7 +32,6 @@ func.func @omp_threads_affinity(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -
         }
         omp.yield
       }
-      omp.terminator
     }
     omp.terminator
   }
@@ -43,9 +42,7 @@ func.func @omp_threads_affinity(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -
 // CHECK:           [[CST_8_:%.+]] = arith.constant 8 : i32
 // CHECK:           omp.parallel num_threads([[CST_8_]] : i32) proc_bind(spread) {
 }
-
 // -----
-
 func.func @omp_threads(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -> (memref<16x8x128xf32> {onnx.name = "y"}) {
   %c32 = arith.constant 32 : index
   %c0 = arith.constant 0 : index
@@ -76,7 +73,6 @@ func.func @omp_threads(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -> (memref
         }
         omp.yield
       }
-      omp.terminator
     }
     omp.terminator
   }
@@ -120,7 +116,6 @@ func.func @omp_affinity(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -> (memre
         }
         omp.yield
       }
-      omp.terminator
     }
     omp.terminator
   }
@@ -162,7 +157,6 @@ func.func @omp_normal(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -> (memref<
         }
         omp.yield
       }
-      omp.terminator
     }
     omp.terminator
   }
@@ -171,5 +165,4 @@ func.func @omp_normal(%arg0: memref<16x8x128xf32> {onnx.name = "x"}) -> (memref<
 // CHECK-LABEL:  func.func @omp_normal
 // CHECK-SAME:   ([[PARAM_0_:%.+]]: memref<16x8x128xf32> {onnx.name = "x"}) -> (memref<16x8x128xf32> {onnx.name = "y"}) {
 // CHECK:           omp.parallel {
-}
-
+}

utils/clone-mlir.sh

@@ -1,3 +1,3 @@
 git clone -n https://github.com/llvm/llvm-project.git
 # Check out a specific branch that is known to work with ONNX-MLIR.
-cd llvm-project && git checkout 00128a20eec27246719d73ba427bf821883b00b4 && cd ..
+cd llvm-project && git checkout 01d233ff403823389f8480897e41aea84ecbb3d3 && cd ..