Add upper bound validation for DQ block_size in MatMulNBits fusion

onnxruntime/core/optimizer/qdq_transformer/selectors_actions/qdq_actions.cc

@@ -87,7 +87,13 @@ int64_t GetEffectiveBlockSize(const Node& dq_node, int64_t block_size_for_non_bl
   const auto& dq_attrs = dq_node.GetAttributes();
   const auto bs_iter = dq_attrs.find("block_size");
   if (bs_iter != dq_attrs.end() && bs_iter->second.i() > 0) {
-    return bs_iter->second.i();
+    // Mirror the kMaxBlockSize guard from ComputeEffectiveBlockSize. Without this cap, a
+    // model-supplied block_size >= 2^32 truncates to 0 in the MLAS int32 transpose argument,
+    // causing a divide-by-zero at CreateSession. The selector rejects such models, but guard
+    // here too so any future selector bypass cannot reach MLAS with an unsafe value.
+    constexpr int64_t kMaxBlockSize = 256;
+    const int64_t block_size = bs_iter->second.i();
+    return (block_size <= kMaxBlockSize) ? block_size : kMaxBlockSize;
   }
 
   // Derive K from the weight input shape if available. Shape information may be missing even

onnxruntime/core/optimizer/qdq_transformer/selectors_actions/qdq_selectors.cc

@@ -776,7 +776,10 @@ static bool ValidateDQForMatMulNBits(const Graph& graph, const Node& dq_node) {
     }
 
     auto block_size = block_size_iter->second.i();
-    if (block_size < 16 || ((block_size - 1) & block_size)) {
+    // Must be a power-of-two in [16, 256]. Values beyond 256 would overflow int32 in the
+    // MLAS transpose kernel (static_cast<int>(block_size) == 0 for block_size >= 2^32),
+    // causing a divide-by-zero (SIGFPE) at CreateSession time.
+    if (block_size < 16 || block_size > 256 || ((block_size - 1) & block_size)) {
       return false;
     }