Fix sycl::queue leak in usm_ndarray get_queue()/get_device()

CHANGELOG.md

@@ -19,6 +19,7 @@ This release is compatible with NumPy 2.5.
 * Added `dpnp.tensor.broadcast_shapes` to align with the 2025.12 version of the Python array API [#3009](https://github.com/IntelPython/dpnp/pull/3009)
 * Added support for free-threaded Python builds [gh-3026](https://github.com/IntelPython/dpnp/pull/3026)
 * Added `dpnp.broadcast` class implementation [#2901](https://github.com/IntelPython/dpnp/pull/2901)
+* Added `UsmNDArray_RemoveQueueRef` C API function to release a queue reference obtained from `UsmNDArray_GetQueueRef` [#3042](https://github.com/IntelPython/dpnp/pull/3042)
 
 ### Changed
 
@@ -89,6 +90,7 @@ This release is compatible with NumPy 2.5.
 * Fixed `dpnp.ndarray.view` ignoring the USM element offset of a sliced array, which also caused `dpnp.einsum` to silently return wrong results for a single sliced operand with no summed index [#3037](https://github.com/IntelPython/dpnp/pull/3037)
 * Fixed `dpnp.all` and `dpnp.any` aborting when reducing over an empty axis (e.g. an array with a zero-length dimension) [#3021](https://github.com/IntelPython/dpnp/pull/3021)
 * Released the GIL before the blocking OneMKL DFT calls in the FFT extension [#3040](https://github.com/IntelPython/dpnp/pull/3040)
+* Fixed a per-call `sycl::queue` leak in `usm_ndarray::get_queue()`/`get_device()` [#3042](https://github.com/IntelPython/dpnp/pull/3042)
 
 ### Security
 

dpnp/include/dpnp4pybind11.hpp

@@ -88,6 +88,7 @@ class dpnp_capi
     int (*UsmNDArray_GetElementSize_)(PyUSMArrayObject *);
     int (*UsmNDArray_GetFlags_)(PyUSMArrayObject *);
     DPCTLSyclQueueRef (*UsmNDArray_GetQueueRef_)(PyUSMArrayObject *);
+    void (*UsmNDArray_RemoveQueueRef_)(DPCTLSyclQueueRef);
     py::ssize_t (*UsmNDArray_GetOffset_)(PyUSMArrayObject *);
     PyObject *(*UsmNDArray_GetUSMData_)(PyUSMArrayObject *);
     void (*UsmNDArray_SetWritableFlag_)(PyUSMArrayObject *, int);
@@ -186,8 +187,9 @@ class dpnp_capi
           UsmNDArray_GetNDim_(nullptr), UsmNDArray_GetShape_(nullptr),
           UsmNDArray_GetStrides_(nullptr), UsmNDArray_GetTypenum_(nullptr),
           UsmNDArray_GetElementSize_(nullptr), UsmNDArray_GetFlags_(nullptr),
-          UsmNDArray_GetQueueRef_(nullptr), UsmNDArray_GetOffset_(nullptr),
-          UsmNDArray_GetUSMData_(nullptr), UsmNDArray_SetWritableFlag_(nullptr),
+          UsmNDArray_GetQueueRef_(nullptr), UsmNDArray_RemoveQueueRef_(nullptr),
+          UsmNDArray_GetOffset_(nullptr), UsmNDArray_GetUSMData_(nullptr),
+          UsmNDArray_SetWritableFlag_(nullptr),
           UsmNDArray_MakeSimpleFromMemory_(nullptr),
           UsmNDArray_MakeSimpleFromPtr_(nullptr),
           UsmNDArray_MakeFromPtr_(nullptr), USM_ARRAY_C_CONTIGUOUS_(0),
@@ -215,6 +217,7 @@ class dpnp_capi
         this->UsmNDArray_GetElementSize_ = UsmNDArray_GetElementSize;
         this->UsmNDArray_GetFlags_ = UsmNDArray_GetFlags;
         this->UsmNDArray_GetQueueRef_ = UsmNDArray_GetQueueRef;
+        this->UsmNDArray_RemoveQueueRef_ = UsmNDArray_RemoveQueueRef;
         this->UsmNDArray_GetOffset_ = UsmNDArray_GetOffset;
         this->UsmNDArray_GetUSMData_ = UsmNDArray_GetUSMData;
         this->UsmNDArray_SetWritableFlag_ = UsmNDArray_SetWritableFlag;
@@ -489,17 +492,33 @@ class usm_ndarray : public py::object
         PyUSMArrayObject *raw_ar = usm_array_ptr();
 
         auto const &api = detail::dpnp_capi::get();
+        // UsmNDArray_GetQueueRef_ returns an owning copy; free it through
+        // RemoveQueueRef (dpctl's DPCTLQueue_Delete)
         DPCTLSyclQueueRef QRef = api.UsmNDArray_GetQueueRef_(raw_ar);
-        return *(reinterpret_cast<sycl::queue *>(QRef));
+        auto qref_deleter = [](sycl::queue *q) {
+            detail::dpnp_capi::get().UsmNDArray_RemoveQueueRef_(
+                reinterpret_cast<DPCTLSyclQueueRef>(q));
+        };
+        std::unique_ptr<sycl::queue, decltype(qref_deleter)> q_ptr{
+            reinterpret_cast<sycl::queue *>(QRef), qref_deleter};
+        return *q_ptr;
     }
 
     sycl::device get_device() const
     {
         PyUSMArrayObject *raw_ar = usm_array_ptr();
 
         auto const &api = detail::dpnp_capi::get();
+        // UsmNDArray_GetQueueRef_ returns an owning copy; free it through
+        // RemoveQueueRef (dpctl's DPCTLQueue_Delete)
         DPCTLSyclQueueRef QRef = api.UsmNDArray_GetQueueRef_(raw_ar);
-        return reinterpret_cast<sycl::queue *>(QRef)->get_device();
+        auto qref_deleter = [](sycl::queue *q) {
+            detail::dpnp_capi::get().UsmNDArray_RemoveQueueRef_(
+                reinterpret_cast<DPCTLSyclQueueRef>(q));
+        };
+        std::unique_ptr<sycl::queue, decltype(qref_deleter)> q_ptr{
+            reinterpret_cast<sycl::queue *>(QRef), qref_deleter};
+        return q_ptr->get_device();
     }
 
     int get_typenum() const

dpnp/tensor/_usmarray.pyx

@@ -1850,6 +1850,12 @@ cdef api c_dpctl.DPCTLSyclQueueRef UsmNDArray_GetQueueRef(usm_ndarray arr):
     return arr.get_queue_ref()
 
 
+cdef api void UsmNDArray_RemoveQueueRef(c_dpctl.DPCTLSyclQueueRef QRef):
+    """Delete a DPCTLSyclQueueRef previously returned by
+    UsmNDArray_GetQueueRef"""
+    c_dpctl.DPCTLQueue_Delete(QRef)   # safe on NULL
+
+
 cdef api Py_ssize_t UsmNDArray_GetOffset(usm_ndarray arr):
     """Get offset of zero-index array element from the beginning of the USM
     allocation"""

dpnp/tests/tensor/test_usm_ndarray_capi.py

@@ -229,8 +229,19 @@ def test_pyx_capi_get_queue_ref():
         fn_restype=ctypes.c_void_p,
         fn_argtypes=(ctypes.py_object,),
     )
+    remove_queue_ref_fn = _pyx_capi_fnptr_to_callable(
+        X,
+        "UsmNDArray_RemoveQueueRef",
+        b"void (DPCTLSyclQueueRef)",
+        fn_restype=None,
+        fn_argtypes=(ctypes.c_void_p,),
+    )
     queue_ref = get_queue_ref_fn(X)  # address of a copy, should be unequal
     assert queue_ref != X.sycl_queue.addressof_ref()
+    # UsmNDArray_GetQueueRef returns an owning copy;
+    # UsmNDArray_RemoveQueueRef must free it and be a no-op on NULL
+    remove_queue_ref_fn(queue_ref)
+    remove_queue_ref_fn(None)
 
 
 def test_pyx_capi_make_from_memory():