[Mlir-commits] [mlir] [mlir][vector] Verify non-unit strides on `masked/expand/compress` ops and fix `SparseVectorization` miscompilation (PR #210952)

Federico Bruzzone llvmlistbot at llvm.org
Tue Jul 21 07:04:36 PDT 2026


https://github.com/FedericoBruzzone updated https://github.com/llvm/llvm-project/pull/210952

>From ee815c22e10991ea39025101b8bc71d00558f0a4 Mon Sep 17 00:00:00 2001
From: Federico Bruzzone <federico.bruzzone.i at gmail.com>
Date: Tue, 21 Jul 2026 15:41:28 +0200
Subject: [PATCH 1/2] [mlir][sparse] Avoid vectorizing non-contiguous COO
 coordinate loads

SparseVectorization emits vector.maskedload/maskedstore for direct
loop accesses, assuming consecutive loop indices map to consecutive
memory. This is false for sparse_tensor.coordinates of a level inside
a trailing AoS COO region, whose buffer is interleaved with other
levels -- a real, silent miscompile.

Fall back to a scalar load/store when the stride is statically known
to be non-unit, including cases only derivable from the sparse
tensor's encoding before sparse-tensor-codegen materializes the
concrete stride.

Signed-off-by: Federico Bruzzone <federico.bruzzone.i at gmail.com>
---
 .../Transforms/SparseVectorization.cpp        | 50 +++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp b/mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp
index 23436a68535fc..c60ca523d81f3 100644
--- a/mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp
@@ -54,6 +54,50 @@ static bool isInvariantArg(BlockArgument arg, Block *block) {
   return arg.getOwner() != block;
 }
 
+/// Returns true when `mem`'s most minor dimension has a statically known
+/// non-unit stride.
+///
+/// `genVectorLoad/genVectorStore` assume a contiguous
+/// `vector.maskedload/vector.maskedstore` is safe for consecutive loop
+/// indices, which breaks for a strided view extracting one component out
+/// of an interleaved (AoS) COO coordinate buffer.
+///
+/// Example:
+/// A `compressed(nonunique) + singleton` region stores coordinates as
+/// `[row0, col0, row1, col1, ...]`, so a 2-lane masked load of `col[0:2]`
+/// (offset=1) would read physical offsets {1, 2} = `[col0, row1]` instead
+/// of the intended {1, 3} = `[col0, col1]`: a silent miscompile.
+///
+/// NOTE: A stride that can't be proven non-unit by either means is assumed
+/// safe.
+static bool hasKnownNonUnitStride(Value mem) {
+  // sparse_tensor.coordinates isn't lowered to a concrete strided memref
+  // until sparse-tensor-codegen runs, so at this point its type
+  // still has a dynamic stride even when the true stride is already known
+  // from the tensor's encoding -- hence the special case below instead of
+  // trusting the memref type.
+  if (auto toCoords = mem.getDefiningOp<ToCoordinatesOp>()) {
+    SparseTensorType stt = getSparseTensorType(toCoords.getTensor());
+    Level cooStart = stt.getAoSCOOStart();
+    // A single trailing level (lvlRank - cooStart == 1) is not actually
+    // interleaved with anything else, so it degenerates to a contiguous
+    // buffer.
+    if (toCoords.getLevel() >= cooStart)
+      return stt.getLvlRank() - cooStart != 1;
+    return false;
+  }
+
+  auto memTp = dyn_cast<MemRefType>(mem.getType());
+  if (!memTp)
+    return false;
+  SmallVector<int64_t> strides;
+  int64_t offset;
+  if (failed(memTp.getStridesAndOffset(strides, offset)))
+    return false;
+  return !strides.empty() && !ShapedType::isDynamic(strides.back()) &&
+         strides.back() != 1;
+}
+
 /// Constructs vector type for element type.
 static VectorType vectorType(VL vl, Type etp) {
   return VectorType::get(vl.vectorLength, etp, vl.enableVLAVectorization);
@@ -292,6 +336,8 @@ static bool vectorizeSubscripts(PatternRewriter &rewriter, scf::ForOp forOp,
     if (auto load = cast.getDefiningOp<memref::LoadOp>()) {
       if (!innermost)
         return false;
+      if (hasKnownNonUnitStride(load.getMemRef()))
+        return false;
       if (codegen) {
         SmallVector<Value> idxs2(load.getIndices()); // no need to analyze
         Location loc = forOp.getLoc();
@@ -408,6 +454,8 @@ static bool vectorizeExpr(PatternRewriter &rewriter, scf::ForOp forOp, VL vl,
   // a[lo:hi] = ind[lo:hi], where 'lo' denotes the current index
   // and 'hi = lo + vl - 1'.
   if (auto load = dyn_cast<memref::LoadOp>(def)) {
+    if (hasKnownNonUnitStride(load.getMemRef()))
+      return false;
     auto subs = load.getIndices();
     SmallVector<Value> idxs;
     if (vectorizeSubscripts(rewriter, forOp, vl, subs, codegen, vmask, idxs)) {
@@ -582,6 +630,8 @@ static bool vectorizeStmt(PatternRewriter &rewriter, scf::ForOp forOp, VL vl,
     }
   } else if (auto store = dyn_cast<memref::StoreOp>(last)) {
     // Analyze/vectorize store operation.
+    if (hasKnownNonUnitStride(store.getMemRef()))
+      return false;
     auto subs = store.getIndices();
     SmallVector<Value> idxs;
     Value rhs = store.getValue();

>From 982fc5dcf309734db2d09dc4bf16cb5f313f5630 Mon Sep 17 00:00:00 2001
From: Federico Bruzzone <federico.bruzzone.i at gmail.com>
Date: Tue, 21 Jul 2026 15:41:54 +0200
Subject: [PATCH 2/2] [mlir][vector] Reject non-unit strides on
 masked/expand/compress ops

vector.maskedload/maskedstore/expandload/compressstore lower to LLVM
masked intrinsics that read/write N *consecutive* elements from a
single pointer, but none of them verified the memref's minor-dim
stride, so e.g. `strided<[2]>` verified successfully and silently
miscompiled.

Reject statically-known non-unit strides; a dynamic stride is still
accepted (see linked issue for discussion on why this differs from
vector.load/store). expandload/compressstore also gain the
negative-stride check vector.load/store already have.

Depends on the SparseVectorization fix in the preceding commit: the
sparsifier's vectorizer previously relied on maskedload having no
stride check to (unsafely) vectorize non-contiguous COO coordinate
buffers.

Signed-off-by: Federico Bruzzone <federico.bruzzone.i at gmail.com>
---
 mlir/lib/Dialect/Vector/IR/VectorOps.cpp | 51 ++++++++++++++++
 mlir/test/Dialect/Vector/invalid.mlir    | 77 ++++++++++++++++++++++++
 2 files changed, 128 insertions(+)

diff --git a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
index f37083803a2a1..8b594c378912e 100644
--- a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
+++ b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
@@ -6190,6 +6190,33 @@ static LogicalResult verifyLoadStoreMemRefLayout(Operation *op,
   return success();
 }
 
+/// Verifies that `memRefTy`'s most minor dimension does not have a
+/// statically known non-unit stride; a dynamic (not provably unit) stride
+/// is accepted.
+///
+/// This is more permissive than vector.load/store's stride check: ops such
+/// as vector.maskedload/maskedstore and vector.expandload/compressstore are
+/// also used on memrefs whose most minor dimension is contiguous at runtime
+/// but not provable as such at the type level (e.g., buffers produced by
+/// the sparsifier).
+static LogicalResult verifyNonStaticNonUnitStrideRejected(Operation *op,
+                                                          VectorType vecTy,
+                                                          MemRefType memRefTy) {
+  if (!vecTy.isScalable() &&
+      (vecTy.getRank() == 0 || vecTy.getNumElements() == 1))
+    return success();
+
+  SmallVector<int64_t> strides;
+  int64_t offset;
+  if (failed(memRefTy.getStridesAndOffset(strides, offset)))
+    return success();
+
+  if (!strides.empty() && !ShapedType::isDynamic(strides.back()) &&
+      strides.back() != 1)
+    return op->emitOpError("most minor memref dim must have unit stride");
+  return success();
+}
+
 LogicalResult vector::LoadOp::verify() {
   VectorType resVecTy = getVectorType();
   MemRefType memRefTy = getMemRefType();
@@ -6299,6 +6326,9 @@ LogicalResult MaskedLoadOp::verify() {
   VectorType resVType = getVectorType();
   MemRefType memType = getMemRefType();
 
+  if (failed(verifyNonStaticNonUnitStrideRejected(*this, resVType, memType)))
+    return failure();
+
   // Negative strides are not supported on vector.maskedload. The lowering to
   // LLVM emits arithmetic operations (e.g., GEP, mul) with nuw flags that
   // assume non-negative strides to avoid undefined behavior.
@@ -6365,6 +6395,9 @@ LogicalResult MaskedStoreOp::verify() {
   VectorType valueVType = getVectorType();
   MemRefType memType = getMemRefType();
 
+  if (failed(verifyNonStaticNonUnitStrideRejected(*this, valueVType, memType)))
+    return failure();
+
   // Negative strides are not supported on vector.maskedstore. The lowering to
   // LLVM emits arithmetic operations (e.g., GEP, mul) with nuw flags that
   // assume non-negative strides to avoid undefined behavior.
@@ -6648,6 +6681,15 @@ LogicalResult ExpandLoadOp::verify() {
   VectorType resVType = getVectorType();
   MemRefType memType = getMemRefType();
 
+  if (failed(verifyNonStaticNonUnitStrideRejected(*this, resVType, memType)))
+    return failure();
+
+  // Negative strides are not supported on vector.expandload. The lowering to
+  // LLVM emits arithmetic operations (e.g., GEP, mul) with nuw flags that
+  // assume non-negative strides to avoid undefined behavior.
+  if (memref::hasNegativeStaticStride(memType))
+    return emitOpError("memref strides must be non-negative");
+
   if (failed(
           verifyElementTypesMatch(*this, memType, resVType, "base", "result")))
     return failure();
@@ -6702,6 +6744,15 @@ LogicalResult CompressStoreOp::verify() {
   VectorType valueVType = getVectorType();
   MemRefType memType = getMemRefType();
 
+  if (failed(verifyNonStaticNonUnitStrideRejected(*this, valueVType, memType)))
+    return failure();
+
+  // Negative strides are not supported on vector.compressstore. The lowering
+  // to LLVM emits arithmetic operations (e.g., GEP, mul) with nuw flags that
+  // assume non-negative strides to avoid undefined behavior.
+  if (memref::hasNegativeStaticStride(memType))
+    return emitOpError("memref strides must be non-negative");
+
   if (failed(verifyElementTypesMatch(*this, memType, valueVType, "base",
                                      "valueToStore")))
     return failure();
diff --git a/mlir/test/Dialect/Vector/invalid.mlir b/mlir/test/Dialect/Vector/invalid.mlir
index aaa55cface958..9ec17024bc1d3 100644
--- a/mlir/test/Dialect/Vector/invalid.mlir
+++ b/mlir/test/Dialect/Vector/invalid.mlir
@@ -1422,6 +1422,15 @@ func.func @maskedload_negative_stride(%src: memref<100x100xf32, strided<[-100, 1
 
 // -----
 
+func.func @maskedload_non_unit_stride(%src: memref<?xi8, strided<[2], offset: ?>>, %mask: vector<8xi1>, %pass: vector<8xi8>) -> vector<8xi8> {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.maskedload' op most minor memref dim must have unit stride}}
+  %0 = vector.maskedload %src[%c0], %mask, %pass : memref<?xi8, strided<[2], offset: ?>>, vector<8xi1>, vector<8xi8> into vector<8xi8>
+  return %0 : vector<8xi8>
+}
+
+// -----
+
 //===----------------------------------------------------------------------===//
 // vector.maskedstore
 //===----------------------------------------------------------------------===//
@@ -1475,6 +1484,15 @@ func.func @maskedstore_negative_stride(%src: memref<100x100xf32, strided<[-100,
 
 // -----
 
+func.func @maskedstore_non_unit_stride(%src: memref<?xi8, strided<[2], offset: ?>>, %mask: vector<8xi1>, %value: vector<8xi8>) {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.maskedstore' op most minor memref dim must have unit stride}}
+  vector.maskedstore %src[%c0], %mask, %value : memref<?xi8, strided<[2], offset: ?>>, vector<8xi1>, vector<8xi8>
+  return
+}
+
+// -----
+
 func.func @gather_from_vector(%base: vector<16xf32>, %indices: vector<16xi32>,
                                 %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
   %c0 = arith.constant 0 : index
@@ -1742,6 +1760,24 @@ func.func @expand_non_power_of_2_alignment(%base: memref<?xf32>, %mask: vector<1
 
 // -----
 
+func.func @expandload_non_unit_stride(%src: memref<?xi8, strided<[2], offset: ?>>, %mask: vector<8xi1>, %pass_thru: vector<8xi8>) -> vector<8xi8> {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.expandload' op most minor memref dim must have unit stride}}
+  %0 = vector.expandload %src[%c0], %mask, %pass_thru : memref<?xi8, strided<[2], offset: ?>>, vector<8xi1>, vector<8xi8> into vector<8xi8>
+  return %0 : vector<8xi8>
+}
+
+// -----
+
+func.func @expandload_negative_stride(%src: memref<100x100xf32, strided<[-100, 1]>>, %mask: vector<8xi1>, %pass_thru: vector<8xf32>) -> vector<8xf32> {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.expandload' op memref strides must be non-negative}}
+  %0 = vector.expandload %src[%c0, %c0], %mask, %pass_thru : memref<100x100xf32, strided<[-100, 1]>>, vector<8xi1>, vector<8xf32> into vector<8xf32>
+  return %0 : vector<8xf32>
+}
+
+// -----
+
 func.func @compress_base_type_mismatch(%base: memref<?xf64>, %mask: vector<16xi1>, %value: vector<16xf32>) {
   %c0 = arith.constant 0 : index
   // expected-error at +1 {{'vector.compressstore' op base element type ('f64') does not match valueToStore element type ('f32')}}
@@ -1796,6 +1832,24 @@ func.func @compress_non_power_of_2_alignment(%base: memref<?xf32>, %mask: vector
 
 // -----
 
+func.func @compressstore_non_unit_stride(%src: memref<?xi8, strided<[2], offset: ?>>, %mask: vector<8xi1>, %value: vector<8xi8>) {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.compressstore' op most minor memref dim must have unit stride}}
+  vector.compressstore %src[%c0], %mask, %value : memref<?xi8, strided<[2], offset: ?>>, vector<8xi1>, vector<8xi8>
+  return
+}
+
+// -----
+
+func.func @compressstore_negative_stride(%src: memref<100x100xf32, strided<[-100, 1]>>, %mask: vector<8xi1>, %value: vector<8xf32>) {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.compressstore' op memref strides must be non-negative}}
+  vector.compressstore %src[%c0, %c0], %mask, %value : memref<100x100xf32, strided<[-100, 1]>>, vector<8xi1>, vector<8xf32>
+  return
+}
+
+// -----
+
 func.func @scan_reduction_dim_constraint(%arg0: vector<2x3xi32>, %arg1: vector<3xi32>) -> vector<3xi32> {
   // expected-error at +1 {{'vector.scan' op reduction dimension 5 has to be less than 2}}
   %0:2 = vector.scan <add>, %arg0, %arg1 {inclusive = true, reduction_dim = 5} :
@@ -2179,6 +2233,19 @@ func.func @load_non_unit_stride(%src : memref<?xi8, strided<[2], offset: ?>>) {
 
 // -----
 
+// Unlike vector.maskedload/maskedstore/expandload/compressstore, a dynamic
+// (unprovable) stride is rejected here too: vector.load/store require a
+// statically known unit stride, with no exception for strides that merely
+// aren't provably non-unit.
+func.func @load_dynamic_stride(%src : memref<?xi8, strided<[?], offset: ?>>) {
+  %c0 = arith.constant 0 : index
+  // expected-error @+1 {{'vector.load' op most minor memref dim must have unit stride}}
+  %0 = vector.load %src[%c0] : memref<?xi8, strided<[?], offset: ?>>, vector<16xi8>
+  return
+}
+
+// -----
+
 //===----------------------------------------------------------------------===//
 // vector.store
 //===----------------------------------------------------------------------===//
@@ -2215,6 +2282,16 @@ func.func @store_non_unit_stride(%src : memref<?xi8, strided<[2], offset:?>>,%va
 
 // -----
 
+// See load_dynamic_stride above: vector.store also rejects a dynamic stride,
+// unlike vector.maskedstore.
+func.func @store_dynamic_stride(%src : memref<?xi8, strided<[?], offset: ?>>, %val : vector<16xi8>, %c0: index) {
+  // expected-error @below {{'vector.store' op most minor memref dim must have unit stride}}
+  vector.store %val, %src[%c0] : memref<?xi8, strided<[?], offset: ?>>, vector<16xi8>
+  return
+}
+
+// -----
+
 func.func @store_negative_stride(%src: memref<100x100xf32, strided<[-100, 1]>>, %val: vector<4xf32>) {
   // expected-error @+2 {{'vector.store' op memref strides must be non-negative}}
   %c0 = arith.constant 0 : index



More information about the Mlir-commits mailing list