Fix unsafe usage in cudnn & clippy warnings/errors (#310)

src/cudnn/result.rs

@@ -458,6 +458,9 @@ pub unsafe fn convolution_forward(
         .result()
 }
 
+/// # Safety
+/// Make sure the handle is valid, all data are associated with the handle, and no pointers are null
+/// unless explicitly accepted by the underlying apis.
 #[allow(clippy::too_many_arguments)]
 pub unsafe fn convolution_bias_activation_forward(
     handle: sys::cudnnHandle_t,
@@ -846,7 +849,10 @@ pub fn create_pooling_descriptor() -> Result<sys::cudnnPoolingDescriptor_t, Cudn
     }
 }
 
-pub fn set_pooling_descriptor(
+/// # Safety
+/// Make sure the handle is valid, all data are associated with the handle, and no pointers are null
+/// unless explicitly accepted by the underlying apis.
+pub unsafe fn set_pooling_descriptor(
     desc: sys::cudnnPoolingDescriptor_t,
     mode: sys::cudnnPoolingMode_t,
     nan_propagation: sys::cudnnNanPropagation_t,
@@ -855,22 +861,24 @@ pub fn set_pooling_descriptor(
     pads: &[std::ffi::c_int],
     strides: &[std::ffi::c_int],
 ) -> Result<(), CudnnError> {
-    unsafe {
-        lib()
-            .cudnnSetPoolingNdDescriptor(
-                desc,
-                mode,
-                nan_propagation,
-                nb_dims,
-                window_dims.as_ptr(),
-                pads.as_ptr(),
-                strides.as_ptr(),
-            )
-            .result()
-    }
+    lib()
+        .cudnnSetPoolingNdDescriptor(
+            desc,
+            mode,
+            nan_propagation,
+            nb_dims,
+            window_dims.as_ptr(),
+            pads.as_ptr(),
+            strides.as_ptr(),
+        )
+        .result()
 }
 
-pub fn pooling_forward(
+/// # Safety
+/// Make sure the handle is valid, all data are associated with the handle, and no pointers are null
+/// unless explicitly accepted by the underlying apis.
+#[allow(clippy::too_many_arguments)]
+pub unsafe fn pooling_forward(
     handle: sys::cudnnHandle_t,
     pooling_desc: sys::cudnnPoolingDescriptor_t,
     alpha: *const ::core::ffi::c_void,
@@ -880,11 +888,9 @@ pub fn pooling_forward(
     y_desc: sys::cudnnTensorDescriptor_t,
     y: *mut ::core::ffi::c_void,
 ) -> Result<(), CudnnError> {
-    unsafe {
-        lib()
-            .cudnnPoolingForward(handle, pooling_desc, alpha, x_desc, x, beta, y_desc, y)
-            .result()
-    }
+    lib()
+        .cudnnPoolingForward(handle, pooling_desc, alpha, x_desc, x, beta, y_desc, y)
+        .result()
 }
 
 pub fn create_activation_descriptor() -> Result<sys::cudnnActivationDescriptor_t, CudnnError> {
@@ -897,20 +903,25 @@ pub fn create_activation_descriptor() -> Result<sys::cudnnActivationDescriptor_t
     }
 }
 
-pub fn set_activation_descriptor(
+/// # Safety
+/// Make sure the handle is valid, all data are associated with the handle, and no pointers are null
+/// unless explicitly accepted by the underlying apis.
+pub unsafe fn set_activation_descriptor(
     desc: sys::cudnnActivationDescriptor_t,
     mode: sys::cudnnActivationMode_t,
     nan_propagation: sys::cudnnNanPropagation_t,
     coef: f64,
 ) -> Result<(), CudnnError> {
-    unsafe {
-        lib()
-            .cudnnSetActivationDescriptor(desc, mode, nan_propagation, coef)
-            .result()
-    }
+    lib()
+        .cudnnSetActivationDescriptor(desc, mode, nan_propagation, coef)
+        .result()
 }
 
-pub fn activation_forward(
+/// # Safety
+/// Make sure the handle is valid, all data are associated with the handle, and no pointers are null
+/// unless explicitly accepted by the underlying apis.
+#[allow(clippy::too_many_arguments)]
+pub unsafe fn activation_forward(
     handle: sys::cudnnHandle_t,
     activation_desc: sys::cudnnActivationDescriptor_t,
     alpha: *const ::core::ffi::c_void,
@@ -920,9 +931,7 @@ pub fn activation_forward(
     y_desc: sys::cudnnTensorDescriptor_t,
     y: *mut ::core::ffi::c_void,
 ) -> Result<(), CudnnError> {
-    unsafe {
-        lib()
-            .cudnnActivationForward(handle, activation_desc, alpha, x_desc, x, beta, y_desc, y)
-            .result()
-    }
+    lib()
+        .cudnnActivationForward(handle, activation_desc, alpha, x_desc, x, beta, y_desc, y)
+        .result()
 }

src/cudnn/safe/activation.rs

@@ -25,7 +25,7 @@ impl Cudnn {
             handle: self.clone(),
             marker: PhantomData,
         };
-        result::set_activation_descriptor(desc.desc, mode, nan_propagation, coef)?;
+        unsafe { result::set_activation_descriptor(desc.desc, mode, nan_propagation, coef) }?;
         Ok(desc)
     }
 }
@@ -39,7 +39,7 @@ pub struct ActivationForward<'a, A: CudnnDataType, X: CudnnDataType, Y: CudnnDat
     pub y: &'a TensorDescriptor<Y>,
 }
 
-impl<'a, A, X, Y> ActivationForward<'a, A, X, Y>
+impl<A, X, Y> ActivationForward<'_, A, X, Y>
 where
     A: CudnnDataType,
     X: CudnnDataType,

src/cudnn/safe/conv.rs

@@ -208,7 +208,7 @@ pub struct ConvForward<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType>
 #[deprecated(note = "use ConvForward instead. This will be removed in future versions")]
 pub type Conv2dForward<'a, X, C, Y> = ConvForward<'a, X, C, Y>;
 
-impl<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvForward<'a, X, C, Y> {
+impl<X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvForward<'_, X, C, Y> {
     /// Picks the fastest algorithm from all available cuDNN algorithms based on cudnn heuristics.
     pub fn pick_algorithm(&self) -> Result<sys::cudnnConvolutionFwdAlgo_t, CudnnError> {
         const NUM_ALGOS: usize = 8;
@@ -322,7 +322,7 @@ pub struct ConvBackwardData<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnData
 #[deprecated(note = "use ConvBackwardData instead. This will be removed in future versions")]
 pub type Conv2dBackwardData<'a, X, C, Y> = ConvBackwardData<'a, X, C, Y>;
 
-impl<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvBackwardData<'a, X, C, Y> {
+impl<X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvBackwardData<'_, X, C, Y> {
     /// Picks the fastest algorithm from all available cuDNN algorithms based on cudnn heuristics.
     pub fn pick_algorithm(&self) -> Result<sys::cudnnConvolutionBwdDataAlgo_t, CudnnError> {
         const NUM_ALGOS: usize = 6;
@@ -436,7 +436,7 @@ pub struct ConvBackwardFilter<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnDa
 #[deprecated(note = "use ConvBackwardFilter instead. This will be removed in future versions")]
 pub type Conv2dBackwardFilter<'a, X, C, Y> = ConvBackwardFilter<'a, X, C, Y>;
 
-impl<'a, X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvBackwardFilter<'a, X, C, Y> {
+impl<X: CudnnDataType, C: CudnnDataType, Y: CudnnDataType> ConvBackwardFilter<'_, X, C, Y> {
     /// Picks the fastest algorithm from all available cuDNN algorithms based on cudnn heuristics.
     pub fn pick_algorithm(&self) -> Result<sys::cudnnConvolutionBwdFilterAlgo_t, CudnnError> {
         const NUM_ALGOS: usize = 7;
@@ -557,7 +557,7 @@ pub struct ConvBiasActivationForward<
     pub y: &'a TensorDescriptor<Y>,
 }
 
-impl<'a, X, C, A, Y> ConvBiasActivationForward<'a, X, C, A, Y>
+impl<X, C, A, Y> ConvBiasActivationForward<'_, X, C, A, Y>
 where
     X: CudnnDataType,
     C: CudnnDataType,
@@ -598,6 +598,7 @@ where
     /// # Safety
     /// The src/filter/y arguments must match the data type/layout specified in the
     /// descriptors in `self.
+    #[allow(clippy::too_many_arguments)]
     pub unsafe fn launch<Workspace, Src, Filter, Dst>(
         &self,
         algo: sys::cudnnConvolutionFwdAlgo_t,

src/cudnn/safe/pooling.rs

@@ -31,15 +31,17 @@ impl Cudnn {
             marker: PhantomData,
         };
 
-        result::set_pooling_descriptor(
-            desc.desc,
-            mode,
-            nan_propagation,
-            window.len() as std::ffi::c_int,
-            window,
-            pads,
-            strides,
-        )?;
+        unsafe {
+            result::set_pooling_descriptor(
+                desc.desc,
+                mode,
+                nan_propagation,
+                window.len() as std::ffi::c_int,
+                window,
+                pads,
+                strides,
+            )
+        }?;
 
         Ok(desc)
     }
@@ -53,7 +55,7 @@ pub struct PoolingForward<'a, P, X, Y> {
     pub y: &'a TensorDescriptor<Y>,
 }
 
-impl<'a, P, X, Y> PoolingForward<'a, P, X, Y>
+impl<P, X, Y> PoolingForward<'_, P, X, Y>
 where
     P: CudnnDataType,
     X: CudnnDataType,

src/cudnn/safe/reduce.rs

@@ -103,7 +103,7 @@ pub struct ReduceTensor<'a, T: CudnnDataType, Idx> {
     pub c: &'a TensorDescriptor<T>,
 }
 
-impl<'a, T: CudnnDataType> ReduceTensor<'a, T, FlatIndices> {
+impl<T: CudnnDataType> ReduceTensor<'_, T, FlatIndices> {
     /// Get's the size of the indices tensor required for this operation.
     ///
     /// See [nvidia docs](https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnGetReductionIndicesSize).
@@ -119,7 +119,7 @@ impl<'a, T: CudnnDataType> ReduceTensor<'a, T, FlatIndices> {
     }
 }
 
-impl<'a, T: CudnnDataType, Idx> ReduceTensor<'a, T, Idx> {
+impl<T: CudnnDataType, Idx> ReduceTensor<'_, T, Idx> {
     /// Gets the size of the workspace for this operation.
     ///
     /// See [nvidia docs](https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnGetReductionWorkspaceSize)
@@ -135,7 +135,7 @@ impl<'a, T: CudnnDataType, Idx> ReduceTensor<'a, T, Idx> {
     }
 }
 
-impl<'a, T: CudnnDataType> ReduceTensor<'a, T, FlatIndices> {
+impl<T: CudnnDataType> ReduceTensor<'_, T, FlatIndices> {
     /// Launches the operation with indices.
     ///
     /// # Safety
@@ -172,7 +172,7 @@ impl<'a, T: CudnnDataType> ReduceTensor<'a, T, FlatIndices> {
     }
 }
 
-impl<'a, T: CudnnDataType> ReduceTensor<'a, T, NoIndices> {
+impl<T: CudnnDataType> ReduceTensor<'_, T, NoIndices> {
     /// Launches the operation with no indices.
     ///
     /// # Safety

src/nccl/safe.rs

@@ -108,7 +108,6 @@ impl Comm {
     ///     let mut slice_receive = dev.alloc_zeros::<f32>(n).unwrap();
     ///     comm.all_reduce(&slice, &mut slice_receive, &ReduceOp::Sum)
     ///         .unwrap();
-
     /// });
     /// group_start().unwrap();
     /// ```
@@ -170,9 +169,7 @@ impl Comm {
     /// let mut slice_receive = dev.alloc_zeros::<f32>(n).unwrap();
     /// comm.all_reduce(&slice, &mut slice_receive, &ReduceOp::Sum)
     ///     .unwrap();
-
     /// let out = dev.dtoh_sync_copy(&slice_receive).unwrap();
-
     /// assert_eq!(out, vec![(n_devices * (n_devices + 1)) as f32 / 2.0; n]);
     /// ```
     pub fn from_rank(

src/runtime/result.rs

@@ -994,8 +994,8 @@ pub mod external_memory {
             size,
             ..Default::default()
         };
-       lib()
-           .cudaImportExternalMemory(external_memory.as_mut_ptr(), &handle_description)
+        lib()
+            .cudaImportExternalMemory(external_memory.as_mut_ptr(), &handle_description)
             .result()?;
         Ok(external_memory.assume_init())
     }