[mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V #151738

YixingZhang007 · 2025-08-01T17:30:44Z

This PR enforces that SPIR-V instructions only allow fixed-size vectors (no scalable vectors) by updating SPIRV_VectorOf to use FixedVectorOfLengthAndType instead of VectorOfLengthAndType. The affected MLIR tests are also updated accordingly.

github-actions · 2025-08-01T17:31:03Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2025-08-01T17:31:35Z

@llvm/pr-subscribers-mlir

Author: None (YixingZhang007)

Changes

This PR enforces that SPIR-V instructions only allow fixed-size vectors (no scalable vectors) by updating SPIRV_VectorOf to use FixedVectorOfLengthAndType instead of VectorOfLengthAndType. The affected MLIR tests are also updated accordingly.

Patch is 30.33 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/151738.diff

11 Files Affected:

(modified) mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir (+12-12)
(modified) mlir/test/Dialect/SPIRV/IR/bit-ops.mlir (+3-3)
(modified) mlir/test/Dialect/SPIRV/IR/gl-ops.mlir (+19-19)
(modified) mlir/test/Dialect/SPIRV/IR/group-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/image-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir (+2-2)
(modified) mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir (+5-5)
(modified) mlir/test/Dialect/SPIRV/IR/logical-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/non-uniform-ops.mlir (+8-8)
(modified) mlir/test/Dialect/SPIRV/IR/ocl-ops.mlir (+6-6)

diff --git a/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td b/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
index 9c74cff0d14f1..37ee85b04f1eb 100644
--- a/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
+++ b/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
@@ -4277,7 +4277,7 @@ class SPIRV_MatrixOfType<list<Type> allowedTypes> :
     "Matrix">;
 
 class SPIRV_VectorOf<Type type> :
-    VectorOfLengthAndType<[2, 3, 4, 8, 16], [type]>;
+    FixedVectorOfLengthAndType<[2, 3, 4, 8, 16], [type]>;
 
 class SPIRV_ScalarOrVectorOf<Type type> :
     AnyTypeOf<[type, SPIRV_VectorOf<type>]>;
diff --git a/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir b/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
index 3adafc15c79f6..c703274bda579 100644
--- a/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
@@ -13,7 +13,7 @@ func.func @fadd_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fadd_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FAdd %arg, %arg : bf16
   return %0 : bf16
 }
@@ -33,7 +33,7 @@ func.func @fdiv_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fdiv_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FDiv %arg, %arg : bf16
   return %0 : bf16
 }
@@ -53,7 +53,7 @@ func.func @fmod_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fmod_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMod %arg, %arg : bf16
   return %0 : bf16
 }
@@ -79,7 +79,7 @@ func.func @fmul_vector(%arg: vector<4xf32>) -> vector<4xf32> {
 // -----
 
 func.func @fmul_i32(%arg: i32) -> i32 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : i32
   return %0 : i32
 }
@@ -87,7 +87,7 @@ func.func @fmul_i32(%arg: i32) -> i32 {
 // -----
 
 func.func @fmul_bf16(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : bf16
   return %0 : bf16
 }
@@ -95,7 +95,7 @@ func.func @fmul_bf16(%arg: bf16) -> bf16 {
 // -----
 
 func.func @fmul_bf16_vector(%arg: vector<4xbf16>) -> vector<4xbf16> {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : vector<4xbf16>
   return %0 : vector<4xbf16>
 }
@@ -103,7 +103,7 @@ func.func @fmul_bf16_vector(%arg: vector<4xbf16>) -> vector<4xbf16> {
 // -----
 
 func.func @fmul_tensor(%arg: tensor<4xf32>) -> tensor<4xf32> {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : tensor<4xf32>
   return %0 : tensor<4xf32>
 }
@@ -123,7 +123,7 @@ func.func @fnegate_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fnegate_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FNegate %arg : bf16
   return %0 : bf16
 }
@@ -143,7 +143,7 @@ func.func @frem_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @frem_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FRem %arg, %arg : bf16
   return %0 : bf16
 }
@@ -163,7 +163,7 @@ func.func @fsub_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fsub_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FSub %arg, %arg : bf16
   return %0 : bf16
 }
@@ -348,7 +348,7 @@ func.func @dot(%arg0: vector<4xf32>, %arg1: vector<4xf32>) -> f16 {
 // -----
 
 func.func @dot(%arg0: vector<4xi32>, %arg1: vector<4xi32>) -> i32 {
-  // expected-error @+1 {{'spirv.Dot' op operand #0 must be vector of 16/32/64-bit float or BFloat16 values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.Dot' op operand #0 must be fixed-length vector of 16/32/64-bit float or BFloat16 values of length 2/3/4/8/16}}
   %0 = spirv.Dot %arg0, %arg1 : vector<4xi32> -> i32
   return %0 : i32
 }
@@ -558,7 +558,7 @@ func.func @vector_times_scalar(%vector: vector<4xf32>, %scalar: f32) -> vector<3
 // -----
 
 func.func @vector_bf16_times_scalar_bf16(%vector: vector<4xbf16>, %scalar: bf16) -> vector<4xbf16> {
-  // expected-error @+1 {{op operand #0 must be vector of 16/32/64-bit float values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be vector of 16/32/64-bit float values of length 2/3/4}}
   %0 = spirv.VectorTimesScalar %vector, %scalar : (vector<4xbf16>, bf16) -> vector<4xbf16>
   return %0 : vector<4xbf16>
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir b/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
index f3f0ebf60f468..4bdac198a1e8f 100644
--- a/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
@@ -137,7 +137,7 @@ func.func @bitwise_or_all_ones_vector(%arg: vector<3xi8>) -> vector<3xi8> {
 // -----
 
 func.func @bitwise_or_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseOr %arg0, %arg1 : f16
   return %0 : f16
 }
@@ -165,7 +165,7 @@ func.func @bitwise_xor_vector(%arg: vector<4xi32>) -> vector<4xi32> {
 // -----
 
 func.func @bitwise_xor_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseXor %arg0, %arg1 : f16
   return %0 : f16
 }
@@ -274,7 +274,7 @@ func.func @bitwise_and_zext_vector(%arg: vector<2xi8>) -> vector<2xi32> {
 // -----
 
 func.func @bitwise_and_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseAnd %arg0, %arg1 : f16
   return %0 : f16
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir b/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
index 5c5d94c40e573..fd8a2ffbbddf9 100644
--- a/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
@@ -19,7 +19,7 @@ func.func @expvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @exp(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %2 = spirv.GL.Exp %arg0 : i32
   return
 }
@@ -27,7 +27,7 @@ func.func @exp(%arg0 : i32) -> () {
 // -----
 
 func.func @exp(%arg0 : vector<5xf32>) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values of length 2/3/4}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values of length 2/3/4}}
   %2 = spirv.GL.Exp %arg0 : vector<5xf32>
   return
 }
@@ -51,7 +51,7 @@ func.func @exp(%arg0 : i32) -> () {
 // -----
 
 func.func @exp_bf16(%arg0 : bf16) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values of length 2/3/4}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values of length 2/3/4}}
   %2 = spirv.GL.Exp %arg0 : bf16
   return
 }
@@ -101,7 +101,7 @@ func.func @iminmax(%arg0: i32, %arg1: i32) {
 // -----
 
 func.func @fmaxminbf16vec(%arg0 : vector<3xbf16>, %arg1 : vector<3xbf16>) {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %1 = spirv.GL.FMax %arg0, %arg1 : vector<3xbf16>
   %2 = spirv.GL.FMin %arg0, %arg1 : vector<3xbf16>
   return
@@ -499,7 +499,7 @@ func.func @frexp_struct_mismatch_type(%arg0 : f32) -> () {
 // -----
 
 func.func @frexp_struct_wrong_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %2 = spirv.GL.FrexpStruct %arg0 : i32 -> !spirv.struct<(i32, i32)>
   return
 }
@@ -614,7 +614,7 @@ func.func @findimsb_vector_i64(%arg0 : vector<3xi64>) -> () {
 // -----
 
 func.func @findimsb_error_scalar_float(%arg0 : f32) -> () {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4/8/1}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4/8/1}}
   %2 = spirv.GL.FindILsb %arg0 : f32
   return
 }
@@ -640,7 +640,7 @@ func.func @findsmsb_vector(%arg0 : vector<3xi32>) -> () {
 // -----
 
 func.func @findsmsb_error_scalar_i64(%arg0 : i64) -> () {
-  // expected-error @+1 {{operand #0 must be Int32 or vector of Int32}}
+  // expected-error @+1 {{operand #0 must be Int32 or fixed-length vector of Int32}}
   %2 = spirv.GL.FindSMsb %arg0 : i64
   return
 }
@@ -666,7 +666,7 @@ func.func @findumsb_vector(%arg0 : vector<3xi32>) -> () {
 // -----
 
 func.func @findumsb(%arg0 : i64) -> () {
-  // expected-error @+1 {{operand #0 must be Int32 or vector of Int32}}
+  // expected-error @+1 {{operand #0 must be Int32 or fixed-length vector of Int32}}
   %2 = spirv.GL.FindUMsb %arg0 : i64
   return
 }
@@ -692,7 +692,7 @@ func.func @distance_vector(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @distance_invalid_type(%arg0 : i32, %arg1 : i32) {
-  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16}}
   %0 = spirv.GL.Distance %arg0, %arg1 : i32, i32 -> f32
   return
 }
@@ -708,7 +708,7 @@ func.func @distance_arg_mismatch(%arg0 : vector<3xf32>, %arg1 : vector<4xf32>) {
 // -----
 
 func.func @distance_invalid_vector_size(%arg0 : vector<5xf32>, %arg1 : vector<5xf32>) {
-  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16}}
   %0 = spirv.GL.Distance %arg0, %arg1 : vector<5xf32>, vector<5xf32> -> f32
   return
 }
@@ -736,7 +736,7 @@ func.func @cross(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @cross_invalid_type(%arg0 : vector<3xi32>, %arg1 : vector<3xi32>) {
-  // expected-error @+1 {{'spirv.GL.Cross' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
+  // expected-error @+1 {{'spirv.GL.Cross' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
   %0 = spirv.GL.Cross %arg0, %arg1 : vector<3xi32>
   return
 }
@@ -762,7 +762,7 @@ func.func @normalize_vector(%arg0 : vector<3xf32>) {
 // -----
 
 func.func @normalize_invalid_type(%arg0 : i32) {
-  // expected-error @+1 {{'spirv.GL.Normalize' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Normalize' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Normalize %arg0 : i32
   return
 }
@@ -788,7 +788,7 @@ func.func @reflect_vector(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @reflect_invalid_type(%arg0 : i32, %arg1 : i32) {
-  // expected-error @+1 {{'spirv.GL.Reflect' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Reflect' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Reflect %arg0, %arg1 : i32
   return
 }
@@ -814,7 +814,7 @@ func.func @fractvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @fract_invalid_type(%arg0 : i32) {
-  // expected-error @+1 {{'spirv.GL.Fract' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Fract' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Fract %arg0 : i32
   return
 }
@@ -840,7 +840,7 @@ func.func @log2vec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @log2_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Log2 %arg0 : i32
   return
 }
@@ -866,7 +866,7 @@ func.func @tanhvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @tanh_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Tanh %arg0 : i32
   return
 }
@@ -892,7 +892,7 @@ func.func @exp2vec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @exp2_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Exp2 %arg0 : i32
   return
 }
@@ -1022,7 +1022,7 @@ func.func @lengthvec(%arg0 : vector<3xf32>) -> () {
 // -----
 
 func.func @length_i32_in(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'i32'}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'i32'}}
   %0 = spirv.GL.Length %arg0 : i32 -> f32
   return
 }
@@ -1038,7 +1038,7 @@ func.func @length_f16_in(%arg0 : f16) -> () {
 // -----
 
 func.func @length_i32vec_in(%arg0 : vector<3xi32>) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
   %0 = spirv.GL.Length %arg0 : vector<3xi32> -> f32
   return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/group-ops.mlir b/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
index d9957ad804161..d7a4a6d92fcd3 100644
--- a/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
@@ -49,7 +49,7 @@ func.func @group_broadcast_negative_scope(%value: f32, %localid: vector<3xi32> )
 // -----
 
 func.func @group_broadcast_negative_locid_dtype(%value: f32, %localid: vector<3xf32> ) -> f32 {
-  // expected-error @+1 {{operand #1 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values}}
+  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values}}
   %0 = spirv.GroupBroadcast <Subgroup> %value, %localid : f32, vector<3xf32>
   return %0: f32
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/image-ops.mlir b/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
index d3aaef7ebdef6..320a8fa360a5f 100644
--- a/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
@@ -349,7 +349,7 @@ func.func @image_fetch_2d_result(%arg0: !spirv.image<f32, Dim2D, NoDepth, NonArr
 // -----
 
 func.func @image_fetch_float_coords(%arg0: !spirv.image<f32, Dim2D, NoDepth, NonArrayed, SingleSampled, NeedSampler, Rgba8>, %arg1: vector<2xf32>) -> () {
-  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4/8/16, but got 'vector<2xf32>'}}
+  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4/8/16, but got 'vector<2xf32>'}}
   %0 = spirv.ImageFetch %arg0, %arg1 : !spirv.image<f32, Dim2D, NoDepth, NonArrayed, SingleSampled, NeedSampler, Rgba8>, vector<2xf32> -> vector<2xf32>
   spirv.Return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir b/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
index bb15d018a6c44..22352da07cf13 100644
--- a/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
@@ -21,7 +21,7 @@ spirv.func @f32_to_bf16_vec(%arg0 : vector<2xf32>) "None" {
 // -----
 
 spirv.func @f32_to_bf16_unsupported(%arg0 : f64) "None" {
-  // expected-error @+1 {{operand #0 must be Float32 or vector of Float32 values of length 2/3/4/8/16, but got}}
+  // expected-error @+1 {{operand #0 must be Float32 or fixed-length vector of Float32 values of length 2/3/4/8/16, but got}}
   %0 = spirv.INTEL.ConvertFToBF16 %arg0 : f64 to i16
   spirv.Return
 }
@@ -57,7 +57,7 @@ spirv.func @bf16_to_f32_vec(%arg0 : vector<2xi16>) "None" {
 // -----
 
 spirv.func @bf16_to_f32_unsupported(%arg0 : i16) "None" {
-  // expected-error @+1 {{result #0 must be Float32 or vector of Float32 values of length 2/3/4/8/16, but got}}
+  // expected-error @+1 {{result #0 must be Float32 or fixed-length vector of Float32 values of length 2/3/4/8/16, but got}}
   %0 = spirv.INTEL.ConvertBF16ToF %arg0 : i16 to f16
   spirv.Return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir b/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
index 61a35b7c991ba..491c7a7758ce1 100644
--- a/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
@@ -583,7 +583,7 @@ spirv.func @matrix_times_scalar(%a: !spirv.coopmatrix<2x2xf32, Workgroup, Matrix
 // These binary arithmetic instructions do not support coop matrix operands.
 
 spirv.func @fmod(%a: !spirv.coopmatrix<2x2xf32, Subgroup, MatrixA>, %b: !spirv.coopmatrix<2x2xf32, Subgroup, MatrixA>) "None" {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector o...
[truncated]

llvmbot · 2025-08-01T17:31:35Z

@llvm/pr-subscribers-mlir-spirv

Author: None (YixingZhang007)

Changes

This PR enforces that SPIR-V instructions only allow fixed-size vectors (no scalable vectors) by updating SPIRV_VectorOf to use FixedVectorOfLengthAndType instead of VectorOfLengthAndType. The affected MLIR tests are also updated accordingly.

Patch is 30.33 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/151738.diff

11 Files Affected:

(modified) mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir (+12-12)
(modified) mlir/test/Dialect/SPIRV/IR/bit-ops.mlir (+3-3)
(modified) mlir/test/Dialect/SPIRV/IR/gl-ops.mlir (+19-19)
(modified) mlir/test/Dialect/SPIRV/IR/group-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/image-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir (+2-2)
(modified) mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir (+5-5)
(modified) mlir/test/Dialect/SPIRV/IR/logical-ops.mlir (+1-1)
(modified) mlir/test/Dialect/SPIRV/IR/non-uniform-ops.mlir (+8-8)
(modified) mlir/test/Dialect/SPIRV/IR/ocl-ops.mlir (+6-6)

diff --git a/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td b/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
index 9c74cff0d14f1..37ee85b04f1eb 100644
--- a/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
+++ b/mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td
@@ -4277,7 +4277,7 @@ class SPIRV_MatrixOfType<list<Type> allowedTypes> :
     "Matrix">;
 
 class SPIRV_VectorOf<Type type> :
-    VectorOfLengthAndType<[2, 3, 4, 8, 16], [type]>;
+    FixedVectorOfLengthAndType<[2, 3, 4, 8, 16], [type]>;
 
 class SPIRV_ScalarOrVectorOf<Type type> :
     AnyTypeOf<[type, SPIRV_VectorOf<type>]>;
diff --git a/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir b/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
index 3adafc15c79f6..c703274bda579 100644
--- a/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/arithmetic-ops.mlir
@@ -13,7 +13,7 @@ func.func @fadd_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fadd_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FAdd %arg, %arg : bf16
   return %0 : bf16
 }
@@ -33,7 +33,7 @@ func.func @fdiv_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fdiv_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FDiv %arg, %arg : bf16
   return %0 : bf16
 }
@@ -53,7 +53,7 @@ func.func @fmod_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fmod_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMod %arg, %arg : bf16
   return %0 : bf16
 }
@@ -79,7 +79,7 @@ func.func @fmul_vector(%arg: vector<4xf32>) -> vector<4xf32> {
 // -----
 
 func.func @fmul_i32(%arg: i32) -> i32 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : i32
   return %0 : i32
 }
@@ -87,7 +87,7 @@ func.func @fmul_i32(%arg: i32) -> i32 {
 // -----
 
 func.func @fmul_bf16(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : bf16
   return %0 : bf16
 }
@@ -95,7 +95,7 @@ func.func @fmul_bf16(%arg: bf16) -> bf16 {
 // -----
 
 func.func @fmul_bf16_vector(%arg: vector<4xbf16>) -> vector<4xbf16> {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : vector<4xbf16>
   return %0 : vector<4xbf16>
 }
@@ -103,7 +103,7 @@ func.func @fmul_bf16_vector(%arg: vector<4xbf16>) -> vector<4xbf16> {
 // -----
 
 func.func @fmul_tensor(%arg: tensor<4xf32>) -> tensor<4xf32> {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FMul %arg, %arg : tensor<4xf32>
   return %0 : tensor<4xf32>
 }
@@ -123,7 +123,7 @@ func.func @fnegate_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fnegate_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FNegate %arg : bf16
   return %0 : bf16
 }
@@ -143,7 +143,7 @@ func.func @frem_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @frem_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FRem %arg, %arg : bf16
   return %0 : bf16
 }
@@ -163,7 +163,7 @@ func.func @fsub_scalar(%arg: f32) -> f32 {
 // -----
 
 func.func @fsub_bf16_scalar(%arg: bf16) -> bf16 {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.FSub %arg, %arg : bf16
   return %0 : bf16
 }
@@ -348,7 +348,7 @@ func.func @dot(%arg0: vector<4xf32>, %arg1: vector<4xf32>) -> f16 {
 // -----
 
 func.func @dot(%arg0: vector<4xi32>, %arg1: vector<4xi32>) -> i32 {
-  // expected-error @+1 {{'spirv.Dot' op operand #0 must be vector of 16/32/64-bit float or BFloat16 values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.Dot' op operand #0 must be fixed-length vector of 16/32/64-bit float or BFloat16 values of length 2/3/4/8/16}}
   %0 = spirv.Dot %arg0, %arg1 : vector<4xi32> -> i32
   return %0 : i32
 }
@@ -558,7 +558,7 @@ func.func @vector_times_scalar(%vector: vector<4xf32>, %scalar: f32) -> vector<3
 // -----
 
 func.func @vector_bf16_times_scalar_bf16(%vector: vector<4xbf16>, %scalar: bf16) -> vector<4xbf16> {
-  // expected-error @+1 {{op operand #0 must be vector of 16/32/64-bit float values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be vector of 16/32/64-bit float values of length 2/3/4}}
   %0 = spirv.VectorTimesScalar %vector, %scalar : (vector<4xbf16>, bf16) -> vector<4xbf16>
   return %0 : vector<4xbf16>
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir b/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
index f3f0ebf60f468..4bdac198a1e8f 100644
--- a/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/bit-ops.mlir
@@ -137,7 +137,7 @@ func.func @bitwise_or_all_ones_vector(%arg: vector<3xi8>) -> vector<3xi8> {
 // -----
 
 func.func @bitwise_or_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseOr %arg0, %arg1 : f16
   return %0 : f16
 }
@@ -165,7 +165,7 @@ func.func @bitwise_xor_vector(%arg: vector<4xi32>) -> vector<4xi32> {
 // -----
 
 func.func @bitwise_xor_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseXor %arg0, %arg1 : f16
   return %0 : f16
 }
@@ -274,7 +274,7 @@ func.func @bitwise_and_zext_vector(%arg: vector<2xi8>) -> vector<2xi32> {
 // -----
 
 func.func @bitwise_and_float(%arg0: f16, %arg1: f16) -> f16 {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4}}
   %0 = spirv.BitwiseAnd %arg0, %arg1 : f16
   return %0 : f16
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir b/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
index 5c5d94c40e573..fd8a2ffbbddf9 100644
--- a/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/gl-ops.mlir
@@ -19,7 +19,7 @@ func.func @expvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @exp(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %2 = spirv.GL.Exp %arg0 : i32
   return
 }
@@ -27,7 +27,7 @@ func.func @exp(%arg0 : i32) -> () {
 // -----
 
 func.func @exp(%arg0 : vector<5xf32>) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values of length 2/3/4}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values of length 2/3/4}}
   %2 = spirv.GL.Exp %arg0 : vector<5xf32>
   return
 }
@@ -51,7 +51,7 @@ func.func @exp(%arg0 : i32) -> () {
 // -----
 
 func.func @exp_bf16(%arg0 : bf16) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values of length 2/3/4}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values of length 2/3/4}}
   %2 = spirv.GL.Exp %arg0 : bf16
   return
 }
@@ -101,7 +101,7 @@ func.func @iminmax(%arg0: i32, %arg1: i32) {
 // -----
 
 func.func @fmaxminbf16vec(%arg0 : vector<3xbf16>, %arg1 : vector<3xbf16>) {
-  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %1 = spirv.GL.FMax %arg0, %arg1 : vector<3xbf16>
   %2 = spirv.GL.FMin %arg0, %arg1 : vector<3xbf16>
   return
@@ -499,7 +499,7 @@ func.func @frexp_struct_mismatch_type(%arg0 : f32) -> () {
 // -----
 
 func.func @frexp_struct_wrong_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %2 = spirv.GL.FrexpStruct %arg0 : i32 -> !spirv.struct<(i32, i32)>
   return
 }
@@ -614,7 +614,7 @@ func.func @findimsb_vector_i64(%arg0 : vector<3xi64>) -> () {
 // -----
 
 func.func @findimsb_error_scalar_float(%arg0 : f32) -> () {
-  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4/8/1}}
+  // expected-error @+1 {{operand #0 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4/8/1}}
   %2 = spirv.GL.FindILsb %arg0 : f32
   return
 }
@@ -640,7 +640,7 @@ func.func @findsmsb_vector(%arg0 : vector<3xi32>) -> () {
 // -----
 
 func.func @findsmsb_error_scalar_i64(%arg0 : i64) -> () {
-  // expected-error @+1 {{operand #0 must be Int32 or vector of Int32}}
+  // expected-error @+1 {{operand #0 must be Int32 or fixed-length vector of Int32}}
   %2 = spirv.GL.FindSMsb %arg0 : i64
   return
 }
@@ -666,7 +666,7 @@ func.func @findumsb_vector(%arg0 : vector<3xi32>) -> () {
 // -----
 
 func.func @findumsb(%arg0 : i64) -> () {
-  // expected-error @+1 {{operand #0 must be Int32 or vector of Int32}}
+  // expected-error @+1 {{operand #0 must be Int32 or fixed-length vector of Int32}}
   %2 = spirv.GL.FindUMsb %arg0 : i64
   return
 }
@@ -692,7 +692,7 @@ func.func @distance_vector(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @distance_invalid_type(%arg0 : i32, %arg1 : i32) {
-  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16}}
   %0 = spirv.GL.Distance %arg0, %arg1 : i32, i32 -> f32
   return
 }
@@ -708,7 +708,7 @@ func.func @distance_arg_mismatch(%arg0 : vector<3xf32>, %arg1 : vector<4xf32>) {
 // -----
 
 func.func @distance_invalid_vector_size(%arg0 : vector<5xf32>, %arg1 : vector<5xf32>) {
-  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16}}
+  // expected-error @+1 {{'spirv.GL.Distance' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16}}
   %0 = spirv.GL.Distance %arg0, %arg1 : vector<5xf32>, vector<5xf32> -> f32
   return
 }
@@ -736,7 +736,7 @@ func.func @cross(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @cross_invalid_type(%arg0 : vector<3xi32>, %arg1 : vector<3xi32>) {
-  // expected-error @+1 {{'spirv.GL.Cross' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
+  // expected-error @+1 {{'spirv.GL.Cross' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
   %0 = spirv.GL.Cross %arg0, %arg1 : vector<3xi32>
   return
 }
@@ -762,7 +762,7 @@ func.func @normalize_vector(%arg0 : vector<3xf32>) {
 // -----
 
 func.func @normalize_invalid_type(%arg0 : i32) {
-  // expected-error @+1 {{'spirv.GL.Normalize' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Normalize' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Normalize %arg0 : i32
   return
 }
@@ -788,7 +788,7 @@ func.func @reflect_vector(%arg0 : vector<3xf32>, %arg1 : vector<3xf32>) {
 // -----
 
 func.func @reflect_invalid_type(%arg0 : i32, %arg1 : i32) {
-  // expected-error @+1 {{'spirv.GL.Reflect' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Reflect' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Reflect %arg0, %arg1 : i32
   return
 }
@@ -814,7 +814,7 @@ func.func @fractvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @fract_invalid_type(%arg0 : i32) {
-  // expected-error @+1 {{'spirv.GL.Fract' op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values}}
+  // expected-error @+1 {{'spirv.GL.Fract' op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values}}
   %0 = spirv.GL.Fract %arg0 : i32
   return
 }
@@ -840,7 +840,7 @@ func.func @log2vec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @log2_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Log2 %arg0 : i32
   return
 }
@@ -866,7 +866,7 @@ func.func @tanhvec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @tanh_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Tanh %arg0 : i32
   return
 }
@@ -892,7 +892,7 @@ func.func @exp2vec(%arg0 : vector<3xf16>) -> () {
 // -----
 
 func.func @exp2_invalid_type(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32-bit float or vector of 16/32-bit float values}}
+  // expected-error @+1 {{op operand #0 must be 16/32-bit float or fixed-length vector of 16/32-bit float values}}
   %0 = spirv.GL.Exp2 %arg0 : i32
   return
 }
@@ -1022,7 +1022,7 @@ func.func @lengthvec(%arg0 : vector<3xf32>) -> () {
 // -----
 
 func.func @length_i32_in(%arg0 : i32) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'i32'}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'i32'}}
   %0 = spirv.GL.Length %arg0 : i32 -> f32
   return
 }
@@ -1038,7 +1038,7 @@ func.func @length_f16_in(%arg0 : f16) -> () {
 // -----
 
 func.func @length_i32vec_in(%arg0 : vector<3xi32>) -> () {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
+  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or fixed-length vector of 16/32/64-bit float values of length 2/3/4/8/16, but got 'vector<3xi32>'}}
   %0 = spirv.GL.Length %arg0 : vector<3xi32> -> f32
   return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/group-ops.mlir b/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
index d9957ad804161..d7a4a6d92fcd3 100644
--- a/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/group-ops.mlir
@@ -49,7 +49,7 @@ func.func @group_broadcast_negative_scope(%value: f32, %localid: vector<3xi32> )
 // -----
 
 func.func @group_broadcast_negative_locid_dtype(%value: f32, %localid: vector<3xf32> ) -> f32 {
-  // expected-error @+1 {{operand #1 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values}}
+  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values}}
   %0 = spirv.GroupBroadcast <Subgroup> %value, %localid : f32, vector<3xf32>
   return %0: f32
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/image-ops.mlir b/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
index d3aaef7ebdef6..320a8fa360a5f 100644
--- a/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/image-ops.mlir
@@ -349,7 +349,7 @@ func.func @image_fetch_2d_result(%arg0: !spirv.image<f32, Dim2D, NoDepth, NonArr
 // -----
 
 func.func @image_fetch_float_coords(%arg0: !spirv.image<f32, Dim2D, NoDepth, NonArrayed, SingleSampled, NeedSampler, Rgba8>, %arg1: vector<2xf32>) -> () {
-  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4/8/16, but got 'vector<2xf32>'}}
+  // expected-error @+1 {{op operand #1 must be 8/16/32/64-bit integer or fixed-length vector of 8/16/32/64-bit integer values of length 2/3/4/8/16, but got 'vector<2xf32>'}}
   %0 = spirv.ImageFetch %arg0, %arg1 : !spirv.image<f32, Dim2D, NoDepth, NonArrayed, SingleSampled, NeedSampler, Rgba8>, vector<2xf32> -> vector<2xf32>
   spirv.Return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir b/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
index bb15d018a6c44..22352da07cf13 100644
--- a/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/intel-ext-ops.mlir
@@ -21,7 +21,7 @@ spirv.func @f32_to_bf16_vec(%arg0 : vector<2xf32>) "None" {
 // -----
 
 spirv.func @f32_to_bf16_unsupported(%arg0 : f64) "None" {
-  // expected-error @+1 {{operand #0 must be Float32 or vector of Float32 values of length 2/3/4/8/16, but got}}
+  // expected-error @+1 {{operand #0 must be Float32 or fixed-length vector of Float32 values of length 2/3/4/8/16, but got}}
   %0 = spirv.INTEL.ConvertFToBF16 %arg0 : f64 to i16
   spirv.Return
 }
@@ -57,7 +57,7 @@ spirv.func @bf16_to_f32_vec(%arg0 : vector<2xi16>) "None" {
 // -----
 
 spirv.func @bf16_to_f32_unsupported(%arg0 : i16) "None" {
-  // expected-error @+1 {{result #0 must be Float32 or vector of Float32 values of length 2/3/4/8/16, but got}}
+  // expected-error @+1 {{result #0 must be Float32 or fixed-length vector of Float32 values of length 2/3/4/8/16, but got}}
   %0 = spirv.INTEL.ConvertBF16ToF %arg0 : i16 to f16
   spirv.Return
 }
diff --git a/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir b/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
index 61a35b7c991ba..491c7a7758ce1 100644
--- a/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
+++ b/mlir/test/Dialect/SPIRV/IR/khr-cooperative-matrix-ops.mlir
@@ -583,7 +583,7 @@ spirv.func @matrix_times_scalar(%a: !spirv.coopmatrix<2x2xf32, Workgroup, Matrix
 // These binary arithmetic instructions do not support coop matrix operands.
 
 spirv.func @fmod(%a: !spirv.coopmatrix<2x2xf32, Subgroup, MatrixA>, %b: !spirv.coopmatrix<2x2xf32, Subgroup, MatrixA>) "None" {
-  // expected-error @+1 {{op operand #0 must be 16/32/64-bit float or vector o...
[truncated]

kuhar

Thank you!

github-actions · 2025-08-01T18:03:48Z

@YixingZhang007 Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

YixingZhang007 added 2 commits August 1, 2025 10:17

update SPIRV_VectorOf and some tests, still have more test to change

aca64cf

made changes in more test

52d1fa5

YixingZhang007 requested review from antiagainst and kuhar as code owners August 1, 2025 17:30

llvmbot added mlir:spirv mlir labels Aug 1, 2025

YixingZhang007 changed the title ~~[mlir][spirv] Enforce Fixed-Size Vectors in SPIR-V Instructions~~ [mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V Aug 1, 2025

kuhar approved these changes Aug 1, 2025

View reviewed changes

YixingZhang007 mentioned this pull request Aug 1, 2025

[mlir][spirv] Add OpExtension "SPV_INTEL_tensor_float32_conversion" #151337

Merged

kuhar merged commit afce932 into llvm:main Aug 1, 2025
12 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V #151738

[mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V #151738

Uh oh!

YixingZhang007 commented Aug 1, 2025

Uh oh!

github-actions bot commented Aug 1, 2025

Uh oh!

llvmbot commented Aug 1, 2025

Uh oh!

llvmbot commented Aug 1, 2025

Uh oh!

kuhar left a comment

Uh oh!

Uh oh!

github-actions bot commented Aug 1, 2025

Uh oh!

Uh oh!

[mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V #151738

[mlir][spirv] Enforce Fixed-Size Vectors for SPIR-V #151738

Uh oh!

Conversation

YixingZhang007 commented Aug 1, 2025

Uh oh!

github-actions bot commented Aug 1, 2025

Uh oh!

llvmbot commented Aug 1, 2025

Uh oh!

llvmbot commented Aug 1, 2025

Uh oh!

kuhar left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Aug 1, 2025

Uh oh!

Uh oh!