Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[mlir][emitc] Set default dialect to emitc in ops with block #123036

Merged
merged 1 commit into from
Jan 20, 2025
Merged

[mlir][emitc] Set default dialect to emitc in ops with block #123036

merged 1 commit into from
Jan 20, 2025

Conversation

kchibisov
Copy link
Contributor

This is a follow up to 68a3908 (func: Set default dialect to 'emitc'), but for other instructions with blocks to make it look consistent.

--

I'm not sure whether the example for the emitc.expression should be updated, probably it's fine to be more explicit within the example.

@kchibisov
[mlir][emitc] Set default dialect to emitc in ops with block
007ba24 
This is a follow up to 68a3908 (func: Set default dialect to
'emitc'), but for other instructions with blocks to make it look
consistent.
@llvmbot
Copy link
Member

llvmbot commented Jan 15, 2025

@llvm/pr-subscribers-mlir-emitc

Author: Kirill Chibisov (kchibisov)

Changes

This is a follow up to 68a3908 (func: Set default dialect to 'emitc'), but for other instructions with blocks to make it look consistent.

--

I'm not sure whether the example for the emitc.expression should be updated, probably it's fine to be more explicit within the example.

Patch is 30.54 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/123036.diff

6 Files Affected:

  • (modified) mlir/include/mlir/Dialect/EmitC/IR/EmitC.td (+41-5)
  • (modified) mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir (+33-27)
  • (modified) mlir/test/Conversion/SCFToEmitC/for.mlir (+16-16)
  • (modified) mlir/test/Conversion/SCFToEmitC/if.mlir (+11-11)
  • (modified) mlir/test/Conversion/SCFToEmitC/switch.mlir (+15-15)
  • (modified) mlir/test/Dialect/EmitC/transforms.mlir (+25-25)
diff --git a/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td b/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
index b16f5a8619fe7b..abcc00feb58163 100644
--- a/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
+++ b/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
@@ -386,8 +386,8 @@ def EmitC_DivOp : EmitC_BinaryOp<"div", [CExpression]> {
 }
 
 def EmitC_ExpressionOp : EmitC_Op<"expression",
-      [HasOnlyGraphRegion, SingleBlockImplicitTerminator<"emitc::YieldOp">,
-       NoRegionArguments]> {
+      [HasOnlyGraphRegion, OpAsmOpInterface,
+       SingleBlockImplicitTerminator<"emitc::YieldOp">, NoRegionArguments]> {
   let summary = "Expression operation";
   let description = [{
     The `emitc.expression` operation returns a single SSA value which is yielded by
@@ -453,12 +453,21 @@ def EmitC_ExpressionOp : EmitC_Op<"expression",
       return llvm::any_of(getRegion().front().without_terminator(), predicate);
     };
     Operation *getRootOp();
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 }
 
 def EmitC_ForOp : EmitC_Op<"for",
       [AllTypesMatch<["lowerBound", "upperBound", "step"]>,
-       SingleBlockImplicitTerminator<"emitc::YieldOp">,
+       OpAsmOpInterface, SingleBlockImplicitTerminator<"emitc::YieldOp">,
        RecursiveMemoryEffects]> {
   let summary = "For operation";
   let description = [{
@@ -510,6 +519,15 @@ def EmitC_ForOp : EmitC_Op<"for",
     void setLowerBound(Value bound) { getOperation()->setOperand(0, bound); }
     void setUpperBound(Value bound) { getOperation()->setOperand(1, bound); }
     void setStep(Value step) { getOperation()->setOperand(2, step); }
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 
   let hasCanonicalizer = 1;
@@ -1257,7 +1275,7 @@ def EmitC_YieldOp : EmitC_Op<"yield",
 def EmitC_IfOp : EmitC_Op<"if",
     [DeclareOpInterfaceMethods<RegionBranchOpInterface, [
     "getNumRegionInvocations", "getRegionInvocationBounds",
-    "getEntrySuccessorRegions"]>, SingleBlock,
+    "getEntrySuccessorRegions"]>, OpAsmOpInterface, SingleBlock,
     SingleBlockImplicitTerminator<"emitc::YieldOp">,
     RecursiveMemoryEffects, NoRegionArguments]> {
   let summary = "If-then-else operation";
@@ -1305,6 +1323,15 @@ def EmitC_IfOp : EmitC_Op<"if",
       Block* body = getBody(1);
       return OpBuilder::atBlockEnd(body, listener);
     }
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
   let hasCustomAssemblyFormat = 1;
 }
@@ -1360,7 +1387,7 @@ def EmitC_SubscriptOp : EmitC_Op<"subscript", []> {
 }
 
 def EmitC_SwitchOp : EmitC_Op<"switch", [RecursiveMemoryEffects,
-    SingleBlockImplicitTerminator<"emitc::YieldOp">,
+    OpAsmOpInterface, SingleBlockImplicitTerminator<"emitc::YieldOp">,
     DeclareOpInterfaceMethods<RegionBranchOpInterface,
                               ["getRegionInvocationBounds",
                                "getEntrySuccessorRegions"]>]> {
@@ -1438,6 +1465,15 @@ def EmitC_SwitchOp : EmitC_Op<"switch", [RecursiveMemoryEffects,
 
     /// Get the body of a case region.
     Block &getCaseBlock(unsigned idx);
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 
   let hasVerifier = 1;
diff --git a/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir b/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
index 1728c3a2557e07..1946677da1baba 100644
--- a/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
+++ b/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
@@ -152,11 +152,12 @@ func.func @arith_shift_left(%arg0: i32, %arg1: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[C2]], %[[SizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : ui32
-  // CHECK: %[[SHL:[^ ]*]] = emitc.bitwise_left_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : ui32
-  // CHECK: emitc.yield %[[Ternary]] : ui32
-  // CHECK: emitc.cast %[[ShiftRes]] : ui32 to i32
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : ui32 {
+  // CHECK-NEXT:   %[[SHL:[^ ]*]] = bitwise_left_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : ui32
+  // CHECK-NEXT:   yield %[[Ternary]] : ui32
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[ShiftRes]] : ui32 to i32
   %1 = arith.shli %arg0, %arg1 : i32
   return
 }
@@ -171,21 +172,23 @@ func.func @arith_shift_right(%arg0: i32, %arg1: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32{{.*}}ui32
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[C2]], %[[SizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}ui32
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : ui32
-  // CHECK: %[[SHR:[^ ]*]] = emitc.bitwise_right_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : ui32
-  // CHECK: emitc.yield %[[Ternary]] : ui32
-  // CHECK: emitc.cast %[[ShiftRes]] : ui32 to i32
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : ui32 {
+  // CHECK-NEXT:   %[[SHR:[^ ]*]] = bitwise_right_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : ui32
+  // CHECK-NEXT:   yield %[[Ternary]] : ui32
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[ShiftRes]] : ui32 to i32
   %2 = arith.shrui %arg0, %arg1 : i32
 
   // CHECK-DAG: %[[SC2:[^ ]*]] = emitc.cast %[[ARG1]] : i32 to ui32
   // CHECK-DAG: %[[SSizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32{{.*}}ui32
   // CHECK-DAG: %[[SCmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[SC2]], %[[SSizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[SZero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}i32
-  // CHECK: %[[SShiftRes:[^ ]*]] = emitc.expression : i32
-  // CHECK: %[[SHRSI:[^ ]*]] = emitc.bitwise_right_shift %[[ARG0]], %[[SC2]] : (i32, ui32) -> i32
-  // CHECK: %[[STernary:[^ ]*]] = emitc.conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : i32
-  // CHECK: emitc.yield %[[STernary]] : i32
+  // CHECK:      %[[SShiftRes:[^ ]*]] = emitc.expression : i32 {
+  // CHECK-NEXT:   %[[SHRSI:[^ ]*]] = bitwise_right_shift %[[ARG0]], %[[SC2]] : (i32, ui32) -> i32
+  // CHECK-NEXT:   %[[STernary:[^ ]*]] = conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : i32
+  // CHECK-NEXT:   yield %[[STernary]] : i32
+  // CHECK-NEXT: }
   %3 = arith.shrsi %arg0, %arg1 : i32
 
   return
@@ -206,10 +209,11 @@ func.func @arith_shift_left_index(%amount: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = emitc.mul %[[Byte]], %[[SizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t
-  // CHECK: %[[SHL:[^ ]*]] = emitc.bitwise_left_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : !emitc.size_t
-  // CHECK: emitc.yield %[[Ternary]] : !emitc.size_t
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t {
+  // CHECK-NEXT:   %[[SHL:[^ ]*]] = bitwise_left_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : !emitc.size_t
+  // CHECK-NEXT:   yield %[[Ternary]] : !emitc.size_t
+  // CHECK-NEXT: }
   %1 = arith.shli %cst0, %cast1 : index
   return
 }
@@ -230,10 +234,11 @@ func.func @arith_shift_right_index(%amount: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = emitc.mul %[[Byte]], %[[SizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}!emitc.size_t
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t
-  // CHECK: %[[SHR:[^ ]*]] = emitc.bitwise_right_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : !emitc.size_t
-  // CHECK: emitc.yield %[[Ternary]] : !emitc.size_t
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t {
+  // CHECK-NEXT:   %[[SHR:[^ ]*]] = bitwise_right_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : !emitc.size_t
+  // CHECK-NEXT:   yield %[[Ternary]] : !emitc.size_t
+  // CHECK-NEXT: }
   %2 = arith.shrui %arg0, %arg1 : index
 
   // CHECK-DAG: %[[SC1:[^ ]*]] = emitc.cast %[[C1]] : !emitc.size_t to !emitc.ptrdiff_t
@@ -242,11 +247,12 @@ func.func @arith_shift_right_index(%amount: i32) {
   // CHECK-DAG: %[[SSizeConstant:[^ ]*]] = emitc.mul %[[SByte]], %[[SSizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[SCmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SSizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[SZero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}!emitc.ptrdiff_t
-  // CHECK: %[[SShiftRes:[^ ]*]] = emitc.expression : !emitc.ptrdiff_t
-  // CHECK: %[[SHRSI:[^ ]*]] = emitc.bitwise_right_shift %[[SC1]], %[[AmountIdx]] : (!emitc.ptrdiff_t, !emitc.size_t) -> !emitc.ptrdiff_t
-  // CHECK: %[[STernary:[^ ]*]] = emitc.conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : !emitc.ptrdiff_t
-  // CHECK: emitc.yield %[[STernary]] : !emitc.ptrdiff_t
-  // CHECK: emitc.cast %[[SShiftRes]] : !emitc.ptrdiff_t to !emitc.size_t
+  // CHECK:      %[[SShiftRes:[^ ]*]] = emitc.expression : !emitc.ptrdiff_t {
+  // CHECK-NEXT:   %[[SHRSI:[^ ]*]] = bitwise_right_shift %[[SC1]], %[[AmountIdx]] : (!emitc.ptrdiff_t, !emitc.size_t) -> !emitc.ptrdiff_t
+  // CHECK-NEXT:   %[[STernary:[^ ]*]] = conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : !emitc.ptrdiff_t
+  // CHECK-NEXT:   yield %[[STernary]] : !emitc.ptrdiff_t
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[SShiftRes]] : !emitc.ptrdiff_t to !emitc.size_t
   %3 = arith.shrsi %arg0, %arg1 : index
 
   return
diff --git a/mlir/test/Conversion/SCFToEmitC/for.mlir b/mlir/test/Conversion/SCFToEmitC/for.mlir
index 7f41e636936b84..232a0fb2e82529 100644
--- a/mlir/test/Conversion/SCFToEmitC/for.mlir
+++ b/mlir/test/Conversion/SCFToEmitC/for.mlir
@@ -33,7 +33,7 @@ func.func @simple_std_2_for_loops(%arg0 : index, %arg1 : index, %arg2 : index) {
 // CHECK-NEXT:    %[[VAL_0:.*]] = builtin.unrealized_conversion_cast %[[ARG_0]] : index to !emitc.size_t
 // CHECK-NEXT:    emitc.for %[[VAL_3:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
 // CHECK-NEXT:      %[[VAL_4:.*]] = arith.constant 1 : index
-// CHECK-NEXT:      emitc.for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
+// CHECK-NEXT:      for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
 // CHECK-NEXT:        %[[VAL_6:.*]] = arith.constant 1 : index
 // CHECK-NEXT:      }
 // CHECK-NEXT:    }
@@ -61,11 +61,11 @@ func.func @for_yield(%arg0 : index, %arg1 : index, %arg2 : index) -> (f32, f32)
 // CHECK-NEXT:    emitc.assign %[[VAL_3]] : f32 to %[[VAL_5]] : <f32>
 // CHECK-NEXT:    emitc.assign %[[VAL_4]] : f32 to %[[VAL_6]] : <f32>
 // CHECK-NEXT:    emitc.for %[[VAL_7:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:      %[[VAL_8:.*]] = emitc.load %[[VAL_5]] : <f32>
-// CHECK-NEXT:      %[[VAL_9:.*]] = emitc.load %[[VAL_6]] : <f32>
+// CHECK-NEXT:      %[[VAL_8:.*]] = load %[[VAL_5]] : <f32>
+// CHECK-NEXT:      %[[VAL_9:.*]] = load %[[VAL_6]] : <f32>
 // CHECK-NEXT:      %[[VAL_10:.*]] = arith.addf %[[VAL_8]], %[[VAL_9]] : f32
-// CHECK-NEXT:      emitc.assign %[[VAL_10]] : f32 to %[[VAL_5]] : <f32>
-// CHECK-NEXT:      emitc.assign %[[VAL_10]] : f32 to %[[VAL_6]] : <f32>
+// CHECK-NEXT:      assign %[[VAL_10]] : f32 to %[[VAL_5]] : <f32>
+// CHECK-NEXT:      assign %[[VAL_10]] : f32 to %[[VAL_6]] : <f32>
 // CHECK-NEXT:    }
 // CHECK-NEXT:    %[[VAL_11:.*]] = emitc.load %[[VAL_5]] : <f32>
 // CHECK-NEXT:    %[[VAL_12:.*]] = emitc.load %[[VAL_6]] : <f32>
@@ -92,16 +92,16 @@ func.func @nested_for_yield(%arg0 : index, %arg1 : index, %arg2 : index) -> f32
 // CHECK-NEXT:    %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f32>
 // CHECK-NEXT:    emitc.assign %[[VAL_3]] : f32 to %[[VAL_4]] : <f32>
 // CHECK-NEXT:    emitc.for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:      %[[VAL_6:.*]] = emitc.load %[[VAL_4]] : <f32>
+// CHECK-NEXT:      %[[VAL_6:.*]] = load %[[VAL_4]] : <f32>
 // CHECK-NEXT:      %[[VAL_7:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f32>
-// CHECK-NEXT:      emitc.assign %[[VAL_6]] : f32 to %[[VAL_7]] : <f32>
-// CHECK-NEXT:      emitc.for %[[VAL_8:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:        %[[VAL_9:.*]] = emitc.load %[[VAL_7]] : <f32>  
+// CHECK-NEXT:      assign %[[VAL_6]] : f32 to %[[VAL_7]] : <f32>
+// CHECK-NEXT:      for %[[VAL_8:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
+// CHECK-NEXT:        %[[VAL_9:.*]] = load %[[VAL_7]] : <f32>  
 // CHECK-NEXT:        %[[VAL_10:.*]] = arith.addf %[[VAL_9]], %[[VAL_9]] : f32
-// CHECK-NEXT:        emitc.assign %[[VAL_10]] : f32 to %[[VAL_7]] : <f32>
+// CHECK-NEXT:        assign %[[VAL_10]] : f32 to %[[VAL_7]] : <f32>
 // CHECK-NEXT:      }
-// CHECK-NEXT:      %[[VAL_11:.*]] = emitc.load %[[VAL_7]] : <f32>  
-// CHECK-NEXT:      emitc.assign %[[VAL_11]] : f32 to %[[VAL_4]] : <f32>
+// CHECK-NEXT:      %[[VAL_11:.*]] = load %[[VAL_7]] : <f32>  
+// CHECK-NEXT:      assign %[[VAL_11]] : f32 to %[[VAL_4]] : <f32>
 // CHECK-NEXT:    }
 // CHECK-NEXT:    %[[VAL_12:.*]] = emitc.load %[[VAL_4]] : <f32>  
 // CHECK-NEXT:    return %[[VAL_12]] : f32
@@ -125,8 +125,8 @@ func.func @for_yield_index(%arg0 : index, %arg1 : index, %arg2 : index) -> index
 // CHECK:     %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<!emitc.size_t>
 // CHECK:     emitc.assign %[[VAL_3]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     emitc.for %[[VAL_5:.*]] = %[[VAL_2]] to %[[VAL_1]] step %[[VAL_0]] : !emitc.size_t {
-// CHECK:       %[[V:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
-// CHECK:       emitc.assign %[[V]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
+// CHECK:       %[[V:.*]] = load %[[VAL_4]] : <!emitc.size_t>
+// CHECK:       assign %[[V]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     }
 // CHECK:     %[[V2:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     %[[VAL_8:.*]] = builtin.unrealized_conversion_cast %[[V2]] : !emitc.size_t to index
@@ -153,11 +153,11 @@ func.func @for_yield_update_loop_carried_var(%arg0 : index, %arg1 : index, %arg2
 // CHECK:   %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<!emitc.size_t>
 // CHECK:   emitc.assign %[[VAL_3]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   emitc.for %[[ARG_3:.*]] = %[[VAL_2]] to %[[VAL_1]] step %[[VAL_0]] : !emitc.size_t {
-// CHECK:     %[[V:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
+// CHECK:     %[[V:.*]] = load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     %[[VAL_5:.*]] = builtin.unrealized_conversion_cast %[[V]] : !emitc.size_t to index
 // CHECK:     %[[VAL_6:.*]] = arith.addi %[[VAL_5]], %[[VAL_5]] : index
 // CHECK:     %[[VAL_8:.*]] = builtin.unrealized_conversion_cast %[[VAL_6]] : index to !emitc.size_t
-// CHECK:     emitc.assign %[[VAL_8]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
+// CHECK:     assign %[[VAL_8]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   }
 // CHECK:   %[[V2:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   %[[VAL_9:.*]] = builtin.unrealized_conversion_cast %[[V2]] : !emitc.size_t to index
diff --git a/mlir/test/Conversion/SCFToEmitC/if.mlir b/mlir/test/Conversion/SCFToEmitC/if.mlir
index 7d923785862d85..9acd9b0783d21c 100644
--- a/mlir/test/Conversion/SCFToEmitC/if.mlir
+++ b/mlir/test/Conversion/SCFToEmitC/if.mlir
@@ -10,7 +10,7 @@ func.func @test_if(%arg0: i1, %arg1: f32) {
 // CHECK-SAME:                     %[[VAL_0:.*]]: i1,
 // CHECK-SAME:                     %[[VAL_1:.*]]: f32) {
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_2:.*]] = emitc.call_opaque "func_const"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_2:.*]] = call_opaque "func_const"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    }
 // CHECK-NEXT:    return
 // CHECK-NEXT:  }
@@ -28,9 +28,9 @@ func.func @test_if_else(%arg0: i1, %arg1: f32) {
 // CHECK-SAME:                          %[[VAL_0:.*]]: i1,
 // CHECK-SAME:                          %[[VAL_1:.*]]: f32) {
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_2:.*]] = emitc.call_opaque "func_true"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_2:.*]] = call_opaque "func_true"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    } else {
-// CHECK-NEXT:      %[[VAL_3:.*]] = emitc.call_opaque "func_false"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_3:.*]] = call_opaque "func_false"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    }
 // CHECK-NEXT:    return
 // CHECK-NEXT:  }
@@ -56,15 +56,15 @@ func.func @test_if_yield(%arg0: i1, %arg1: f32) -> (i32, f64) {
 // CHECK-NEXT:    %[[VAL_3:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<i32>
 // CHECK-NEXT:    %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f64>
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_5:.*]] = emitc.call_opaque "func_true_1"(%[[VAL_1]]) : (f32) -> i32
-// CHECK-NEXT:      %[[VAL_6:.*]] = emitc.call_opaque "func_true_2"(%[[VAL_1]]) : (f32) -> f64
-// CHECK-NEXT:      emitc.assign %[[VAL_5]] : i32 to %[[VAL_3]] : <i32>
-// CHECK-NEXT:      emitc.assign %[[VAL_6]] : f64 to %[[VAL_4]] : <f64>
+// CHECK-NEXT:      %[[VAL_5:.*]] = call_opaque "func_true_1"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_6:.*]] = call_opaque "func_true_2"(%[[VAL_1]]) : (f32) -> f64
+// CHECK-NEXT:      assign %[[VAL_5]] : i32 to %[[VAL_3]] : <i32>
+// CHECK-NEXT:      assign %[[VAL_6]] : f64 to %[[VAL_4]] : <f64>
 // CHECK-NEXT:    } else {
-// CHECK-NEXT:      %[[VAL_7:.*]] = emitc.call_opaque "func_false_1"(%[[VAL_1]]) : (f32) -> i32
-// CHECK-NEXT:      %[[VAL_8:.*]] = emitc.call_opaque "func_false_2"(%[[VAL_1]]) : (f32) -> f64
-// CHECK-NEXT:      emitc.assign %[[VAL_7]] : i32 to %[[VAL_3]] : <i32>
-// CHECK-NEXT:      emitc.assign %[[VAL_8]] : f64 to %[[VAL_4]] : <f64>
+// CHECK-NEXT:      %[[VAL_7:.*]] = call_opaque "func_false_1...
[truncated]

@llvmbot
Copy link
Member

llvmbot commented Jan 15, 2025

@llvm/pr-subscribers-mlir

Author: Kirill Chibisov (kchibisov)

Changes

This is a follow up to 68a3908 (func: Set default dialect to 'emitc'), but for other instructions with blocks to make it look consistent.

--

I'm not sure whether the example for the emitc.expression should be updated, probably it's fine to be more explicit within the example.

Patch is 30.54 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/123036.diff

6 Files Affected:

  • (modified) mlir/include/mlir/Dialect/EmitC/IR/EmitC.td (+41-5)
  • (modified) mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir (+33-27)
  • (modified) mlir/test/Conversion/SCFToEmitC/for.mlir (+16-16)
  • (modified) mlir/test/Conversion/SCFToEmitC/if.mlir (+11-11)
  • (modified) mlir/test/Conversion/SCFToEmitC/switch.mlir (+15-15)
  • (modified) mlir/test/Dialect/EmitC/transforms.mlir (+25-25)
diff --git a/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td b/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
index b16f5a8619fe7b..abcc00feb58163 100644
--- a/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
+++ b/mlir/include/mlir/Dialect/EmitC/IR/EmitC.td
@@ -386,8 +386,8 @@ def EmitC_DivOp : EmitC_BinaryOp<"div", [CExpression]> {
 }
 
 def EmitC_ExpressionOp : EmitC_Op<"expression",
-      [HasOnlyGraphRegion, SingleBlockImplicitTerminator<"emitc::YieldOp">,
-       NoRegionArguments]> {
+      [HasOnlyGraphRegion, OpAsmOpInterface,
+       SingleBlockImplicitTerminator<"emitc::YieldOp">, NoRegionArguments]> {
   let summary = "Expression operation";
   let description = [{
     The `emitc.expression` operation returns a single SSA value which is yielded by
@@ -453,12 +453,21 @@ def EmitC_ExpressionOp : EmitC_Op<"expression",
       return llvm::any_of(getRegion().front().without_terminator(), predicate);
     };
     Operation *getRootOp();
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 }
 
 def EmitC_ForOp : EmitC_Op<"for",
       [AllTypesMatch<["lowerBound", "upperBound", "step"]>,
-       SingleBlockImplicitTerminator<"emitc::YieldOp">,
+       OpAsmOpInterface, SingleBlockImplicitTerminator<"emitc::YieldOp">,
        RecursiveMemoryEffects]> {
   let summary = "For operation";
   let description = [{
@@ -510,6 +519,15 @@ def EmitC_ForOp : EmitC_Op<"for",
     void setLowerBound(Value bound) { getOperation()->setOperand(0, bound); }
     void setUpperBound(Value bound) { getOperation()->setOperand(1, bound); }
     void setStep(Value step) { getOperation()->setOperand(2, step); }
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 
   let hasCanonicalizer = 1;
@@ -1257,7 +1275,7 @@ def EmitC_YieldOp : EmitC_Op<"yield",
 def EmitC_IfOp : EmitC_Op<"if",
     [DeclareOpInterfaceMethods<RegionBranchOpInterface, [
     "getNumRegionInvocations", "getRegionInvocationBounds",
-    "getEntrySuccessorRegions"]>, SingleBlock,
+    "getEntrySuccessorRegions"]>, OpAsmOpInterface, SingleBlock,
     SingleBlockImplicitTerminator<"emitc::YieldOp">,
     RecursiveMemoryEffects, NoRegionArguments]> {
   let summary = "If-then-else operation";
@@ -1305,6 +1323,15 @@ def EmitC_IfOp : EmitC_Op<"if",
       Block* body = getBody(1);
       return OpBuilder::atBlockEnd(body, listener);
     }
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
   let hasCustomAssemblyFormat = 1;
 }
@@ -1360,7 +1387,7 @@ def EmitC_SubscriptOp : EmitC_Op<"subscript", []> {
 }
 
 def EmitC_SwitchOp : EmitC_Op<"switch", [RecursiveMemoryEffects,
-    SingleBlockImplicitTerminator<"emitc::YieldOp">,
+    OpAsmOpInterface, SingleBlockImplicitTerminator<"emitc::YieldOp">,
     DeclareOpInterfaceMethods<RegionBranchOpInterface,
                               ["getRegionInvocationBounds",
                                "getEntrySuccessorRegions"]>]> {
@@ -1438,6 +1465,15 @@ def EmitC_SwitchOp : EmitC_Op<"switch", [RecursiveMemoryEffects,
 
     /// Get the body of a case region.
     Block &getCaseBlock(unsigned idx);
+
+    //===------------------------------------------------------------------===//
+    // OpAsmOpInterface Methods
+    //===------------------------------------------------------------------===//
+
+    /// EmitC ops in the body can omit their 'emitc.' prefix in the assembly.
+    static ::llvm::StringRef getDefaultDialect() {
+      return "emitc";
+    }
   }];
 
   let hasVerifier = 1;
diff --git a/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir b/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
index 1728c3a2557e07..1946677da1baba 100644
--- a/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
+++ b/mlir/test/Conversion/ArithToEmitC/arith-to-emitc.mlir
@@ -152,11 +152,12 @@ func.func @arith_shift_left(%arg0: i32, %arg1: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[C2]], %[[SizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : ui32
-  // CHECK: %[[SHL:[^ ]*]] = emitc.bitwise_left_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : ui32
-  // CHECK: emitc.yield %[[Ternary]] : ui32
-  // CHECK: emitc.cast %[[ShiftRes]] : ui32 to i32
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : ui32 {
+  // CHECK-NEXT:   %[[SHL:[^ ]*]] = bitwise_left_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : ui32
+  // CHECK-NEXT:   yield %[[Ternary]] : ui32
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[ShiftRes]] : ui32 to i32
   %1 = arith.shli %arg0, %arg1 : i32
   return
 }
@@ -171,21 +172,23 @@ func.func @arith_shift_right(%arg0: i32, %arg1: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32{{.*}}ui32
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[C2]], %[[SizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}ui32
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : ui32
-  // CHECK: %[[SHR:[^ ]*]] = emitc.bitwise_right_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : ui32
-  // CHECK: emitc.yield %[[Ternary]] : ui32
-  // CHECK: emitc.cast %[[ShiftRes]] : ui32 to i32
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : ui32 {
+  // CHECK-NEXT:   %[[SHR:[^ ]*]] = bitwise_right_shift %[[C1]], %[[C2]] : (ui32, ui32) -> ui32
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : ui32
+  // CHECK-NEXT:   yield %[[Ternary]] : ui32
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[ShiftRes]] : ui32 to i32
   %2 = arith.shrui %arg0, %arg1 : i32
 
   // CHECK-DAG: %[[SC2:[^ ]*]] = emitc.cast %[[ARG1]] : i32 to ui32
   // CHECK-DAG: %[[SSizeConstant:[^ ]*]] = "emitc.constant"{{.*}}value = 32{{.*}}ui32
   // CHECK-DAG: %[[SCmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[SC2]], %[[SSizeConstant]] : (ui32, ui32) -> i1
   // CHECK-DAG: %[[SZero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}i32
-  // CHECK: %[[SShiftRes:[^ ]*]] = emitc.expression : i32
-  // CHECK: %[[SHRSI:[^ ]*]] = emitc.bitwise_right_shift %[[ARG0]], %[[SC2]] : (i32, ui32) -> i32
-  // CHECK: %[[STernary:[^ ]*]] = emitc.conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : i32
-  // CHECK: emitc.yield %[[STernary]] : i32
+  // CHECK:      %[[SShiftRes:[^ ]*]] = emitc.expression : i32 {
+  // CHECK-NEXT:   %[[SHRSI:[^ ]*]] = bitwise_right_shift %[[ARG0]], %[[SC2]] : (i32, ui32) -> i32
+  // CHECK-NEXT:   %[[STernary:[^ ]*]] = conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : i32
+  // CHECK-NEXT:   yield %[[STernary]] : i32
+  // CHECK-NEXT: }
   %3 = arith.shrsi %arg0, %arg1 : i32
 
   return
@@ -206,10 +209,11 @@ func.func @arith_shift_left_index(%amount: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = emitc.mul %[[Byte]], %[[SizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t
-  // CHECK: %[[SHL:[^ ]*]] = emitc.bitwise_left_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : !emitc.size_t
-  // CHECK: emitc.yield %[[Ternary]] : !emitc.size_t
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t {
+  // CHECK-NEXT:   %[[SHL:[^ ]*]] = bitwise_left_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHL]], %[[Zero]] : !emitc.size_t
+  // CHECK-NEXT:   yield %[[Ternary]] : !emitc.size_t
+  // CHECK-NEXT: }
   %1 = arith.shli %cst0, %cast1 : index
   return
 }
@@ -230,10 +234,11 @@ func.func @arith_shift_right_index(%amount: i32) {
   // CHECK-DAG: %[[SizeConstant:[^ ]*]] = emitc.mul %[[Byte]], %[[SizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[CmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[Zero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}!emitc.size_t
-  // CHECK: %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t
-  // CHECK: %[[SHR:[^ ]*]] = emitc.bitwise_right_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
-  // CHECK: %[[Ternary:[^ ]*]] = emitc.conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : !emitc.size_t
-  // CHECK: emitc.yield %[[Ternary]] : !emitc.size_t
+  // CHECK:      %[[ShiftRes:[^ ]*]] = emitc.expression : !emitc.size_t {
+  // CHECK-NEXT:   %[[SHR:[^ ]*]] = bitwise_right_shift %[[C1]], %[[AmountIdx]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
+  // CHECK-NEXT:   %[[Ternary:[^ ]*]] = conditional %[[CmpNoExcess]], %[[SHR]], %[[Zero]] : !emitc.size_t
+  // CHECK-NEXT:   yield %[[Ternary]] : !emitc.size_t
+  // CHECK-NEXT: }
   %2 = arith.shrui %arg0, %arg1 : index
 
   // CHECK-DAG: %[[SC1:[^ ]*]] = emitc.cast %[[C1]] : !emitc.size_t to !emitc.ptrdiff_t
@@ -242,11 +247,12 @@ func.func @arith_shift_right_index(%amount: i32) {
   // CHECK-DAG: %[[SSizeConstant:[^ ]*]] = emitc.mul %[[SByte]], %[[SSizeOf]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
   // CHECK-DAG: %[[SCmpNoExcess:[^ ]*]] = emitc.cmp lt, %[[AmountIdx]], %[[SSizeConstant]] : (!emitc.size_t, !emitc.size_t) -> i1
   // CHECK-DAG: %[[SZero:[^ ]*]] = "emitc.constant"{{.*}}value = 0{{.*}}!emitc.ptrdiff_t
-  // CHECK: %[[SShiftRes:[^ ]*]] = emitc.expression : !emitc.ptrdiff_t
-  // CHECK: %[[SHRSI:[^ ]*]] = emitc.bitwise_right_shift %[[SC1]], %[[AmountIdx]] : (!emitc.ptrdiff_t, !emitc.size_t) -> !emitc.ptrdiff_t
-  // CHECK: %[[STernary:[^ ]*]] = emitc.conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : !emitc.ptrdiff_t
-  // CHECK: emitc.yield %[[STernary]] : !emitc.ptrdiff_t
-  // CHECK: emitc.cast %[[SShiftRes]] : !emitc.ptrdiff_t to !emitc.size_t
+  // CHECK:      %[[SShiftRes:[^ ]*]] = emitc.expression : !emitc.ptrdiff_t {
+  // CHECK-NEXT:   %[[SHRSI:[^ ]*]] = bitwise_right_shift %[[SC1]], %[[AmountIdx]] : (!emitc.ptrdiff_t, !emitc.size_t) -> !emitc.ptrdiff_t
+  // CHECK-NEXT:   %[[STernary:[^ ]*]] = conditional %[[SCmpNoExcess]], %[[SHRSI]], %[[SZero]] : !emitc.ptrdiff_t
+  // CHECK-NEXT:   yield %[[STernary]] : !emitc.ptrdiff_t
+  // CHECK-NEXT: }
+  // CHECK-NEXT: emitc.cast %[[SShiftRes]] : !emitc.ptrdiff_t to !emitc.size_t
   %3 = arith.shrsi %arg0, %arg1 : index
 
   return
diff --git a/mlir/test/Conversion/SCFToEmitC/for.mlir b/mlir/test/Conversion/SCFToEmitC/for.mlir
index 7f41e636936b84..232a0fb2e82529 100644
--- a/mlir/test/Conversion/SCFToEmitC/for.mlir
+++ b/mlir/test/Conversion/SCFToEmitC/for.mlir
@@ -33,7 +33,7 @@ func.func @simple_std_2_for_loops(%arg0 : index, %arg1 : index, %arg2 : index) {
 // CHECK-NEXT:    %[[VAL_0:.*]] = builtin.unrealized_conversion_cast %[[ARG_0]] : index to !emitc.size_t
 // CHECK-NEXT:    emitc.for %[[VAL_3:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
 // CHECK-NEXT:      %[[VAL_4:.*]] = arith.constant 1 : index
-// CHECK-NEXT:      emitc.for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
+// CHECK-NEXT:      for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
 // CHECK-NEXT:        %[[VAL_6:.*]] = arith.constant 1 : index
 // CHECK-NEXT:      }
 // CHECK-NEXT:    }
@@ -61,11 +61,11 @@ func.func @for_yield(%arg0 : index, %arg1 : index, %arg2 : index) -> (f32, f32)
 // CHECK-NEXT:    emitc.assign %[[VAL_3]] : f32 to %[[VAL_5]] : <f32>
 // CHECK-NEXT:    emitc.assign %[[VAL_4]] : f32 to %[[VAL_6]] : <f32>
 // CHECK-NEXT:    emitc.for %[[VAL_7:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:      %[[VAL_8:.*]] = emitc.load %[[VAL_5]] : <f32>
-// CHECK-NEXT:      %[[VAL_9:.*]] = emitc.load %[[VAL_6]] : <f32>
+// CHECK-NEXT:      %[[VAL_8:.*]] = load %[[VAL_5]] : <f32>
+// CHECK-NEXT:      %[[VAL_9:.*]] = load %[[VAL_6]] : <f32>
 // CHECK-NEXT:      %[[VAL_10:.*]] = arith.addf %[[VAL_8]], %[[VAL_9]] : f32
-// CHECK-NEXT:      emitc.assign %[[VAL_10]] : f32 to %[[VAL_5]] : <f32>
-// CHECK-NEXT:      emitc.assign %[[VAL_10]] : f32 to %[[VAL_6]] : <f32>
+// CHECK-NEXT:      assign %[[VAL_10]] : f32 to %[[VAL_5]] : <f32>
+// CHECK-NEXT:      assign %[[VAL_10]] : f32 to %[[VAL_6]] : <f32>
 // CHECK-NEXT:    }
 // CHECK-NEXT:    %[[VAL_11:.*]] = emitc.load %[[VAL_5]] : <f32>
 // CHECK-NEXT:    %[[VAL_12:.*]] = emitc.load %[[VAL_6]] : <f32>
@@ -92,16 +92,16 @@ func.func @nested_for_yield(%arg0 : index, %arg1 : index, %arg2 : index) -> f32
 // CHECK-NEXT:    %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f32>
 // CHECK-NEXT:    emitc.assign %[[VAL_3]] : f32 to %[[VAL_4]] : <f32>
 // CHECK-NEXT:    emitc.for %[[VAL_5:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:      %[[VAL_6:.*]] = emitc.load %[[VAL_4]] : <f32>
+// CHECK-NEXT:      %[[VAL_6:.*]] = load %[[VAL_4]] : <f32>
 // CHECK-NEXT:      %[[VAL_7:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f32>
-// CHECK-NEXT:      emitc.assign %[[VAL_6]] : f32 to %[[VAL_7]] : <f32>
-// CHECK-NEXT:      emitc.for %[[VAL_8:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
-// CHECK-NEXT:        %[[VAL_9:.*]] = emitc.load %[[VAL_7]] : <f32>  
+// CHECK-NEXT:      assign %[[VAL_6]] : f32 to %[[VAL_7]] : <f32>
+// CHECK-NEXT:      for %[[VAL_8:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] : !emitc.size_t {
+// CHECK-NEXT:        %[[VAL_9:.*]] = load %[[VAL_7]] : <f32>  
 // CHECK-NEXT:        %[[VAL_10:.*]] = arith.addf %[[VAL_9]], %[[VAL_9]] : f32
-// CHECK-NEXT:        emitc.assign %[[VAL_10]] : f32 to %[[VAL_7]] : <f32>
+// CHECK-NEXT:        assign %[[VAL_10]] : f32 to %[[VAL_7]] : <f32>
 // CHECK-NEXT:      }
-// CHECK-NEXT:      %[[VAL_11:.*]] = emitc.load %[[VAL_7]] : <f32>  
-// CHECK-NEXT:      emitc.assign %[[VAL_11]] : f32 to %[[VAL_4]] : <f32>
+// CHECK-NEXT:      %[[VAL_11:.*]] = load %[[VAL_7]] : <f32>  
+// CHECK-NEXT:      assign %[[VAL_11]] : f32 to %[[VAL_4]] : <f32>
 // CHECK-NEXT:    }
 // CHECK-NEXT:    %[[VAL_12:.*]] = emitc.load %[[VAL_4]] : <f32>  
 // CHECK-NEXT:    return %[[VAL_12]] : f32
@@ -125,8 +125,8 @@ func.func @for_yield_index(%arg0 : index, %arg1 : index, %arg2 : index) -> index
 // CHECK:     %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<!emitc.size_t>
 // CHECK:     emitc.assign %[[VAL_3]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     emitc.for %[[VAL_5:.*]] = %[[VAL_2]] to %[[VAL_1]] step %[[VAL_0]] : !emitc.size_t {
-// CHECK:       %[[V:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
-// CHECK:       emitc.assign %[[V]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
+// CHECK:       %[[V:.*]] = load %[[VAL_4]] : <!emitc.size_t>
+// CHECK:       assign %[[V]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     }
 // CHECK:     %[[V2:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     %[[VAL_8:.*]] = builtin.unrealized_conversion_cast %[[V2]] : !emitc.size_t to index
@@ -153,11 +153,11 @@ func.func @for_yield_update_loop_carried_var(%arg0 : index, %arg1 : index, %arg2
 // CHECK:   %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<!emitc.size_t>
 // CHECK:   emitc.assign %[[VAL_3]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   emitc.for %[[ARG_3:.*]] = %[[VAL_2]] to %[[VAL_1]] step %[[VAL_0]] : !emitc.size_t {
-// CHECK:     %[[V:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
+// CHECK:     %[[V:.*]] = load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:     %[[VAL_5:.*]] = builtin.unrealized_conversion_cast %[[V]] : !emitc.size_t to index
 // CHECK:     %[[VAL_6:.*]] = arith.addi %[[VAL_5]], %[[VAL_5]] : index
 // CHECK:     %[[VAL_8:.*]] = builtin.unrealized_conversion_cast %[[VAL_6]] : index to !emitc.size_t
-// CHECK:     emitc.assign %[[VAL_8]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
+// CHECK:     assign %[[VAL_8]] : !emitc.size_t to %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   }
 // CHECK:   %[[V2:.*]] = emitc.load %[[VAL_4]] : <!emitc.size_t>
 // CHECK:   %[[VAL_9:.*]] = builtin.unrealized_conversion_cast %[[V2]] : !emitc.size_t to index
diff --git a/mlir/test/Conversion/SCFToEmitC/if.mlir b/mlir/test/Conversion/SCFToEmitC/if.mlir
index 7d923785862d85..9acd9b0783d21c 100644
--- a/mlir/test/Conversion/SCFToEmitC/if.mlir
+++ b/mlir/test/Conversion/SCFToEmitC/if.mlir
@@ -10,7 +10,7 @@ func.func @test_if(%arg0: i1, %arg1: f32) {
 // CHECK-SAME:                     %[[VAL_0:.*]]: i1,
 // CHECK-SAME:                     %[[VAL_1:.*]]: f32) {
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_2:.*]] = emitc.call_opaque "func_const"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_2:.*]] = call_opaque "func_const"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    }
 // CHECK-NEXT:    return
 // CHECK-NEXT:  }
@@ -28,9 +28,9 @@ func.func @test_if_else(%arg0: i1, %arg1: f32) {
 // CHECK-SAME:                          %[[VAL_0:.*]]: i1,
 // CHECK-SAME:                          %[[VAL_1:.*]]: f32) {
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_2:.*]] = emitc.call_opaque "func_true"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_2:.*]] = call_opaque "func_true"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    } else {
-// CHECK-NEXT:      %[[VAL_3:.*]] = emitc.call_opaque "func_false"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_3:.*]] = call_opaque "func_false"(%[[VAL_1]]) : (f32) -> i32
 // CHECK-NEXT:    }
 // CHECK-NEXT:    return
 // CHECK-NEXT:  }
@@ -56,15 +56,15 @@ func.func @test_if_yield(%arg0: i1, %arg1: f32) -> (i32, f64) {
 // CHECK-NEXT:    %[[VAL_3:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<i32>
 // CHECK-NEXT:    %[[VAL_4:.*]] = "emitc.variable"() <{value = #emitc.opaque<"">}> : () -> !emitc.lvalue<f64>
 // CHECK-NEXT:    emitc.if %[[VAL_0]] {
-// CHECK-NEXT:      %[[VAL_5:.*]] = emitc.call_opaque "func_true_1"(%[[VAL_1]]) : (f32) -> i32
-// CHECK-NEXT:      %[[VAL_6:.*]] = emitc.call_opaque "func_true_2"(%[[VAL_1]]) : (f32) -> f64
-// CHECK-NEXT:      emitc.assign %[[VAL_5]] : i32 to %[[VAL_3]] : <i32>
-// CHECK-NEXT:      emitc.assign %[[VAL_6]] : f64 to %[[VAL_4]] : <f64>
+// CHECK-NEXT:      %[[VAL_5:.*]] = call_opaque "func_true_1"(%[[VAL_1]]) : (f32) -> i32
+// CHECK-NEXT:      %[[VAL_6:.*]] = call_opaque "func_true_2"(%[[VAL_1]]) : (f32) -> f64
+// CHECK-NEXT:      assign %[[VAL_5]] : i32 to %[[VAL_3]] : <i32>
+// CHECK-NEXT:      assign %[[VAL_6]] : f64 to %[[VAL_4]] : <f64>
 // CHECK-NEXT:    } else {
-// CHECK-NEXT:      %[[VAL_7:.*]] = emitc.call_opaque "func_false_1"(%[[VAL_1]]) : (f32) -> i32
-// CHECK-NEXT:      %[[VAL_8:.*]] = emitc.call_opaque "func_false_2"(%[[VAL_1]]) : (f32) -> f64
-// CHECK-NEXT:      emitc.assign %[[VAL_7]] : i32 to %[[VAL_3]] : <i32>
-// CHECK-NEXT:      emitc.assign %[[VAL_8]] : f64 to %[[VAL_4]] : <f64>
+// CHECK-NEXT:      %[[VAL_7:.*]] = call_opaque "func_false_1...
[truncated]

mgehre-amd
mgehre-amd approved these changes Jan 15, 2025
View reviewed changes
Copy link
Contributor

@mgehre-amd mgehre-amd left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good!

marbre
marbre approved these changes Jan 15, 2025
View reviewed changes
Copy link
Member

@marbre marbre left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks!

@kchibisov
Copy link
Contributor Author

gentle ping.

@mgehre-amd
Copy link
Contributor

@kchibisov, would you like me to merge on behalf of you?

@kchibisov
Copy link
Contributor Author

kchibisov commented Jan 20, 2025 •
edited
Loading

I can not merge anything, I don't have write access, so yes.

@mgehre-amd mgehre-amd merged commit 977d744 into llvm:main Jan 20, 2025
11 checks passed
@kchibisov kchibisov deleted the emitc-block-asm-printer branch January 20, 2025 13:51
@kchibisov
Copy link
Contributor Author

Thanks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@marbre marbre marbre approved these changes

@mgehre-amd mgehre-amd mgehre-amd approved these changes

Assignees
No one assigned
Labels
mlir:emitc mlir
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@kchibisov @llvmbot @mgehre-amd @marbre