Review fixups

Fixes: #66910 (comment)
Fixes: #66910 (comment)

Author: MacDue

mlir/include/mlir/Dialect/ArmSME/IR/ArmSME.td

@@ -563,7 +563,7 @@ class LLVM_aarch64_sme_read<string direction>
     : ArmSME_IntrOp<"read." # direction, /*overloadedOperands=*/[],
                     [AllShapesMatch<["vector", "pg", "res"]>,
                      AllElementTypesMatch<["vector", "res"]>],
-                    /*numResults*/1, /*overloadedResults*/[0]>,
+                    /*numResults=*/1, /*overloadedResults=*/[0]>,
       Arguments<(ins Arg<SVEVector, "Vector operand">:$vector,
                      Arg<SVEPredicate, "Vector predicate">:$pg,
                      Arg<I32, "Virtual tile ID">,

mlir/test/Target/LLVMIR/arm-sme.mlir

@@ -337,15 +337,6 @@ llvm.func @arm_sme_vector_to_tile_vert(%tileslice : i32,
 
 // -----
 
-llvm.func @prevent_dce.nxv16i8(vector<[16]xi8>)
-llvm.func @prevent_dce.nxv8i16(vector<[8]xi16>)
-llvm.func @prevent_dce.nxv4i32(vector<[4]xi32>)
-llvm.func @prevent_dce.nxv2i64(vector<[2]xi64>)
-llvm.func @prevent_dce.nxv1i128(vector<[1]xi128>)
-llvm.func @prevent_dce.nxv8f16(vector<[8]xf16>)
-llvm.func @prevent_dce.nxv8bf16(vector<[8]xbf16>)
-llvm.func @prevent_dce.nxv4f32(vector<[4]xf32>)
-llvm.func @prevent_dce.nxv2f64(vector<[2]xf64>)
 
 llvm.func @arm_sme_tile_slice_to_vector_horiz(%tileslice : i32,
                                               %nxv16i1 : vector<[16]xi1>,
@@ -366,54 +357,35 @@ llvm.func @arm_sme_tile_slice_to_vector_horiz(%tileslice : i32,
   // CHECK: call <vscale x 16 x i8> @llvm.aarch64.sme.read.horiz.nxv16i8
   %res0 = "arm_sme.intr.read.horiz"(%nxv16i8, %nxv16i1, %tile, %tileslice)
     : (vector<[16]xi8>, vector<[16]xi1>, i32, i32) -> vector<[16]xi8>
-  llvm.call @prevent_dce.nxv16i8(%res0) : (vector<[16]xi8>) -> ()
   // CHECK: call <vscale x 8 x i16> @llvm.aarch64.sme.read.horiz.nxv8i16
   %res1 = "arm_sme.intr.read.horiz"(%nxv8i16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xi16>, vector<[8]xi1>, i32, i32) -> vector<[8]xi16>
-  llvm.call @prevent_dce.nxv8i16(%res1) : (vector<[8]xi16>) -> ()
   // CHECK: call <vscale x 4 x i32> @llvm.aarch64.sme.read.horiz.nxv4i32
   %res2 = "arm_sme.intr.read.horiz"(%nxv4i32, %nxv4i1, %tile, %tileslice)
     : (vector<[4]xi32>, vector<[4]xi1>, i32, i32) -> vector<[4]xi32>
-  llvm.call @prevent_dce.nxv4i32(%res2) : (vector<[4]xi32>) -> ()
   // CHECK: call <vscale x 2 x i64> @llvm.aarch64.sme.read.horiz.nxv2i64
   %res3 = "arm_sme.intr.read.horiz"(%nxv2i64, %nxv2i1, %tile, %tileslice)
     : (vector<[2]xi64>, vector<[2]xi1>, i32, i32) -> vector<[2]xi64>
-  llvm.call @prevent_dce.nxv2i64(%res3) : (vector<[2]xi64>) -> ()
   // CHECK: call <vscale x 1 x i128> @llvm.aarch64.sme.read.horiz.nxv1i128
   %res4 = "arm_sme.intr.read.horiz"(%nxv1i128, %nxv1i1, %tile, %tileslice)
     : (vector<[1]xi128>, vector<[1]xi1>, i32, i32) -> vector<[1]xi128>
-  llvm.call @prevent_dce.nxv1i128(%res4) : (vector<[1]xi128>) -> ()
   // CHECK: call <vscale x 8 x half> @llvm.aarch64.sme.read.horiz.nxv8f16
   %res5 = "arm_sme.intr.read.horiz"(%nxv8f16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xf16>, vector<[8]xi1>, i32, i32) -> vector<[8]xf16>
-  llvm.call @prevent_dce.nxv8f16(%res5) : (vector<[8]xf16>) -> ()
   // CHECK: call <vscale x 8 x bfloat> @llvm.aarch64.sme.read.horiz.nxv8bf16
   %res6 = "arm_sme.intr.read.horiz"(%nxv8bf16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xbf16>, vector<[8]xi1>, i32, i32) -> vector<[8]xbf16>
-  llvm.call @prevent_dce.nxv8bf16(%res6) : (vector<[8]xbf16>) -> ()
   // CHECK: call <vscale x 4 x float> @llvm.aarch64.sme.read.horiz.nxv4f32
   %res7 = "arm_sme.intr.read.horiz"(%nxv4f32, %nxv4i1, %tile, %tileslice)
     : (vector<[4]xf32>, vector<[4]xi1>, i32, i32) -> vector<[4]xf32>
-  llvm.call @prevent_dce.nxv4f32(%res7) : (vector<[4]xf32>) -> ()
   // CHECK: call <vscale x 2 x double> @llvm.aarch64.sme.read.horiz.nxv2f64
   %res8 = "arm_sme.intr.read.horiz"(%nxv2f64, %nxv2i1, %tile, %tileslice)
     : (vector<[2]xf64>, vector<[2]xi1>, i32, i32) -> vector<[2]xf64>
-  llvm.call @prevent_dce.nxv2f64(%res8) : (vector<[2]xf64>) -> ()
   llvm.return
 }
 
 // -----
 
-llvm.func @prevent_dce.nxv16i8(vector<[16]xi8>)
-llvm.func @prevent_dce.nxv8i16(vector<[8]xi16>)
-llvm.func @prevent_dce.nxv4i32(vector<[4]xi32>)
-llvm.func @prevent_dce.nxv2i64(vector<[2]xi64>)
-llvm.func @prevent_dce.nxv1i128(vector<[1]xi128>)
-llvm.func @prevent_dce.nxv8f16(vector<[8]xf16>)
-llvm.func @prevent_dce.nxv8bf16(vector<[8]xbf16>)
-llvm.func @prevent_dce.nxv4f32(vector<[4]xf32>)
-llvm.func @prevent_dce.nxv2f64(vector<[2]xf64>)
-
 llvm.func @arm_sme_tile_slice_to_vector_vert(%tileslice : i32,
                                               %nxv16i1 : vector<[16]xi1>,
                                               %nxv8i1 : vector<[8]xi1>,
@@ -433,38 +405,29 @@ llvm.func @arm_sme_tile_slice_to_vector_vert(%tileslice : i32,
   // CHECK: call <vscale x 16 x i8> @llvm.aarch64.sme.read.vert.nxv16i8
   %res0 = "arm_sme.intr.read.vert"(%nxv16i8, %nxv16i1, %tile, %tileslice)
     : (vector<[16]xi8>, vector<[16]xi1>, i32, i32) -> vector<[16]xi8>
-  llvm.call @prevent_dce.nxv16i8(%res0) : (vector<[16]xi8>) -> ()
   // CHECK: call <vscale x 8 x i16> @llvm.aarch64.sme.read.vert.nxv8i16
   %res1 = "arm_sme.intr.read.vert"(%nxv8i16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xi16>, vector<[8]xi1>, i32, i32) -> vector<[8]xi16>
-  llvm.call @prevent_dce.nxv8i16(%res1) : (vector<[8]xi16>) -> ()
   // CHECK: call <vscale x 4 x i32> @llvm.aarch64.sme.read.vert.nxv4i32
   %res2 = "arm_sme.intr.read.vert"(%nxv4i32, %nxv4i1, %tile, %tileslice)
     : (vector<[4]xi32>, vector<[4]xi1>, i32, i32) -> vector<[4]xi32>
-  llvm.call @prevent_dce.nxv4i32(%res2) : (vector<[4]xi32>) -> ()
   // CHECK: call <vscale x 2 x i64> @llvm.aarch64.sme.read.vert.nxv2i64
   %res3 = "arm_sme.intr.read.vert"(%nxv2i64, %nxv2i1, %tile, %tileslice)
     : (vector<[2]xi64>, vector<[2]xi1>, i32, i32) -> vector<[2]xi64>
-  llvm.call @prevent_dce.nxv2i64(%res3) : (vector<[2]xi64>) -> ()
   // CHECK: call <vscale x 1 x i128> @llvm.aarch64.sme.read.vert.nxv1i128
   %res4 = "arm_sme.intr.read.vert"(%nxv1i128, %nxv1i1, %tile, %tileslice)
     : (vector<[1]xi128>, vector<[1]xi1>, i32, i32) -> vector<[1]xi128>
-  llvm.call @prevent_dce.nxv1i128(%res4) : (vector<[1]xi128>) -> ()
   // CHECK: call <vscale x 8 x half> @llvm.aarch64.sme.read.vert.nxv8f16
   %res5 = "arm_sme.intr.read.vert"(%nxv8f16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xf16>, vector<[8]xi1>, i32, i32) -> vector<[8]xf16>
-  llvm.call @prevent_dce.nxv8f16(%res5) : (vector<[8]xf16>) -> ()
   // CHECK: call <vscale x 8 x bfloat> @llvm.aarch64.sme.read.vert.nxv8bf16
   %res6 = "arm_sme.intr.read.vert"(%nxv8bf16, %nxv8i1, %tile, %tileslice)
     : (vector<[8]xbf16>, vector<[8]xi1>, i32, i32) -> vector<[8]xbf16>
-  llvm.call @prevent_dce.nxv8bf16(%res6) : (vector<[8]xbf16>) -> ()
   // CHECK: call <vscale x 4 x float> @llvm.aarch64.sme.read.vert.nxv4f32
   %res7 = "arm_sme.intr.read.vert"(%nxv4f32, %nxv4i1, %tile, %tileslice)
     : (vector<[4]xf32>, vector<[4]xi1>, i32, i32) -> vector<[4]xf32>
-  llvm.call @prevent_dce.nxv4f32(%res7) : (vector<[4]xf32>) -> ()
   // CHECK: call <vscale x 2 x double> @llvm.aarch64.sme.read.vert.nxv2f64
   %res8 = "arm_sme.intr.read.vert"(%nxv2f64, %nxv2i1, %tile, %tileslice)
     : (vector<[2]xf64>, vector<[2]xi1>, i32, i32) -> vector<[2]xf64>
-  llvm.call @prevent_dce.nxv2f64(%res8) : (vector<[2]xf64>) -> ()
   llvm.return
 }