Fixes #639: Simplify access-permissions-related code:

* Renamed: `access_permissions` -> `permissions` and `access_permissions_t` to `permissions_t`, everwhere
* Dropped the boolean enum in favor of parameter-less some named constructor idioms
* Dropped the static methods in favor of named constructor idioms
* No user code needs to worry about pairs-of-booleans; but they can aggregate-initialize if they really want to

Author: eyalroz

examples/by_api_module/memory_pools.cpp

@@ -130,21 +130,21 @@ void copy_through_pool_allocation(
 	}
 }
 
-cuda::memory::access_permissions_t
+cuda::memory::permissions_t
 try_forbidding_same_device_access(int device_id, cuda::memory::pool_t &pool)
 {
-	cuda::memory::access_permissions_t permissions;
+	cuda::memory::permissions_t permissions;
 	permissions.read = true;
 	permissions.write = false;
 	bool got_expected_exception = false;
 	try {
-		pool.set_access_permissions(cuda::device::get(device_id), permissions);
+		pool.set_permissions(cuda::device::get(device_id), permissions);
 	}
 	catch(std::invalid_argument&) {
 		got_expected_exception = true;
 	}
 	if (not got_expected_exception) {
-		die_("Unexpected success in setting a device's access permissions to a pool using"
+		die_("Unexpected success in setting a device's access get_permissions to a pool using"
 		"that device; it should have failed");
 	}
 	return permissions;
@@ -155,12 +155,12 @@ void try_writing_to_pool_allocation_without_permission(
 	const cuda::stream_t&                stream,
 	cuda::memory::pool_t&                pool,
 	cuda::memory::region_t&              pool_allocated_region,
-	cuda::memory::access_permissions_t&  permissions,
+	cuda::memory::permissions_t&  permissions,
 	cuda::device_t&                      peer)
 {
 	permissions.read = false;
 	permissions.write = false;
-	pool.set_access_permissions(peer, permissions);
+	pool.set_permissions(peer, permissions);
 	std::string str{"hello world"};
 	stream.synchronize();
 	auto stream_on_peer = peer.create_stream(cuda::stream::async);
@@ -184,12 +184,12 @@ void try_writing_to_pool_allocation_without_permission(
 void try_reading_from_pool_allocation_without_permission(
 	cuda::memory::pool_t&                pool,
 	cuda::memory::region_t&              pool_allocated_region,
-	cuda::memory::access_permissions_t&  permissions,
+	cuda::memory::permissions_t&  permissions,
 	cuda::device_t&                      peer)
 {
 	permissions.read = false;
 	permissions.write = false;
-	pool.set_access_permissions(peer, permissions);
+	pool.set_permissions(peer, permissions);
 	auto host_buffer_uptr = std::unique_ptr<char[]>(new char[region_size]); // replace this with make_unique in C++14
 	auto host_buffer = cuda::span<char>{host_buffer_uptr.get(), region_size};
 	std::fill_n(host_buffer.begin(), host_buffer.size()-1, 'a');
@@ -223,7 +223,7 @@ int main(int argc, char** argv)
 	play_with_attributes(pool, stream);
 	copy_through_pool_allocation(stream, pool_allocated_region);
 
-	cuda::memory::access_permissions_t permissions = try_forbidding_same_device_access(device_id, pool);
+	cuda::memory::permissions_t permissions = try_forbidding_same_device_access(device_id, pool);
 
 	auto maybe_peer_id = maybe_get_p2p_peer_id(device_id);
 	if (maybe_peer_id) {

examples/modified_cuda_samples/memMapIPCDrv/child.cpp

@@ -57,8 +57,7 @@ std::vector<memory_mapping_t> import_and_map_allocations(
 		});
 
 	// Retain peer access and map all chunks to mapDevice
-	virtual_mem::set_access_mode(mappings_region, device,
-		cuda::memory::access_permissions_t::read_and_write());
+	virtual_mem::set_permissions(mappings_region, device, cuda::memory::permissions::read_and_write());
 	return mappings;
 }
 

examples/modified_cuda_samples/streamOrderedAllocationIPC/streamOrderedAllocationIPC.cu

@@ -146,9 +146,9 @@ static void childProcess(int index_in_shared_devices)
 		auto shared_pool_handle = shared_pool_handles[i];
 		auto pool_device = cuda::device::get(shm->devices[i]);
 		auto pool = cuda::memory::pool::ipc::import<shared_handle_kind>(pool_device, shared_pool_handle);
-		auto permissions = pool.access_permissions(device);
+		auto permissions = pool.permissions(device);
 		if (not (permissions.read and permissions.write)) {
-			pool.set_access_permissions(device, cuda::memory::read_enabled, cuda::memory::write_enabled);
+			pool.set_permissions(device, cuda::memory::permissions::read_and_write());
 		}
 
 		// Import the allocations from each memory pool

examples/modified_cuda_samples/streamOrderedAllocationP2P/streamOrderedAllocationP2P.cu

@@ -171,8 +171,7 @@ int memPoolP2PCopy()
 
 	auto stream2 = p2pDevices.second.create_stream(cuda::stream::async);
 	auto output_on_device = cuda::span<int>(stream2.enqueue.allocate(nelem * sizeof(int)));
-	memPool.set_access_permissions(p2pDevices.second,
-		cuda::memory::read_enabled, cuda::memory::write_enabled);
+	memPool.set_permissions(p2pDevices.second, cuda::memory::permissions::read_and_write());
 
 	std::cout << "> copyP2PAndScale kernel running ...\n";
 	auto launch_config = cuda::launch_config_builder()

examples/modified_cuda_samples/vectorAddMMAP/vectorAddMMAP.cpp

@@ -167,14 +167,13 @@ setup_virtual_memory(
 	);
 
 #ifndef _MSC_VER
-	virtual_mem::set_access_mode(reserved_range.region(), mapping_devices,
-		cuda::memory::access_permissions_t::read_and_write());
+	virtual_mem::set_permissions(reserved_range.region(), mapping_devices, cuda::memory::permissions::read_and_write());
 #else
 	// MSVC, at least as of 2019, can't handle template-template parameters with variadcs properly;
 	// so let's go manual:
 	for(const auto& mapping_device : mapping_devices) {
-		virtual_mem::set_access_mode(reserved_range.region(), mapping_device,
-			cuda::memory::access_permissions_t::read_and_write());
+		virtual_mem::set_permissions(reserved_range.region(), mapping_device,
+			cuda::memory::permissions::read_and_write());
 	}
 #endif
 

src/cuda/api/memory_pool.hpp

@@ -159,21 +159,21 @@ pool_t wrap(cuda::device::id_t device_id, pool::handle_t handle, bool owning) no
 
 namespace detail_ {
 
-inline access_permissions_t access_permissions(cuda::device::id_t device_id, pool::handle_t pool_handle)
+inline permissions_t get_permissions(cuda::device::id_t device_id, pool::handle_t pool_handle)
 {
 	CUmemAccess_flags access_flags;
 	auto mem_location = pool::detail_::create_mem_location(device_id);
 	auto status = cuMemPoolGetAccess(&access_flags, pool_handle, &mem_location);
 	throw_if_error_lazy(status,
 		"Determining access information for " + cuda::device::detail_::identify(device_id)
 		+ " to " + pool::detail_::identify(pool_handle));
-	return access_permissions_t::from_access_flags(access_flags);
+	return permissions::detail_::from_flags(access_flags);
 }
 
-inline void set_access_permissions(span<cuda::device::id_t> device_ids, pool::handle_t pool_handle, access_permissions_t permissions)
+inline void set_permissions(span<cuda::device::id_t> device_ids, pool::handle_t pool_handle, permissions_t permissions)
 {
 	if (permissions.write and not permissions.read) {
-		throw ::std::invalid_argument("Memory pool access permissions cannot be write-only");
+		throw ::std::invalid_argument("Memory pool access get_permissions cannot be write-only");
 	}
 
 	CUmemAccess_flags flags = permissions.read ?
@@ -192,14 +192,14 @@ inline void set_access_permissions(span<cuda::device::id_t> device_ids, pool::ha
 
 	auto status = cuMemPoolSetAccess(pool_handle, descriptors.data(), descriptors.size());
 	throw_if_error_lazy(status,
-		"Setting access permissions for " + ::std::to_string(descriptors.size())
+		"Setting access get_permissions for " + ::std::to_string(descriptors.size())
 		+ " devices to " + pool::detail_::identify(pool_handle));
 }
 
-inline void set_access_permissions(cuda::device::id_t device_id, pool::handle_t pool_handle, access_permissions_t permissions)
+inline void set_permissions(cuda::device::id_t device_id, pool::handle_t pool_handle, permissions_t permissions)
 {
 	if (permissions.write and not permissions.read) {
-		throw ::std::invalid_argument("Memory pool access permissions cannot be write-only");
+		throw ::std::invalid_argument("Memory pool access get_permissions cannot be write-only");
 	}
 
 	CUmemAccessDesc desc;
@@ -212,16 +212,16 @@ inline void set_access_permissions(cuda::device::id_t device_id, pool::handle_t
 	desc.location = pool::detail_::create_mem_location(device_id);
 	auto status = cuMemPoolSetAccess(pool_handle, &desc, 1);
 	throw_if_error_lazy(status,
-		"Setting access permissions for " + cuda::device::detail_::identify(device_id)
+		"Setting access get_permissions for " + cuda::device::detail_::identify(device_id)
 		+ " to " + pool::detail_::identify(pool_handle));
 }
 
 } // namespace detail_
 
-access_permissions_t access_permissions(const cuda::device_t& device, const pool_t& pool);
-void set_access_permissions(const cuda::device_t& device, const pool_t& pool, access_permissions_t permissions);
+permissions_t get_permissions(const cuda::device_t& device, const pool_t& pool);
+void set_permissions(const cuda::device_t& device, const pool_t& pool, permissions_t permissions);
 template <typename DeviceRange>
-void set_access_permissions(DeviceRange devices, const pool_t& pool_handle, access_permissions_t permissions);
+void get_permissions(DeviceRange devices, const pool_t& pool_handle, permissions_t permissions);
 
 namespace pool {
 
@@ -302,48 +302,38 @@ class pool_t {
 		set_attribute<CU_MEMPOOL_ATTR_RELEASE_THRESHOLD>(threshold);
 	}
 
-	access_permissions_t access_permissions(const cuda::device_t& device)
+	permissions_t permissions(const cuda::device_t& device)
 	{
-		return memory::access_permissions(device, *this);
+		return memory::get_permissions(device, *this);
 	}
 
 	/**
-	 * Set read and write permissions from a device to the allocations from
+	 * Set read and write get_permissions from a device to the allocations from
 	 * this pool
 	 *
-	 * @param device the device the kernels running on which are governed by these permissions
-	 * @param permissions new read and write permissions to use
+	 * @param device the device the kernels running on which are governed by these get_permissions
+	 * @param permissions new read and write get_permissions to use
 	 *
 	 * @note This affects both future _and past_ allocations from this pool.
 	 */
 	///@{
 	/**
 	 * @param device the device the kernels running on which are governed by this new setting
-	 * @param permissions new read and write permissions to use
+	 * @param permissions new read and write get_permissions to use
 	 */
-	void set_access_permissions(const cuda::device_t& device, access_permissions_t permissions)
+	void set_permissions(const cuda::device_t& device, permissions_t permissions)
 	{
-		return memory::set_access_permissions(device, *this, permissions);
-	}
-
-	/**
-	 * @param device the device the kernels running on which are governed by this new setting
-	 * @param read_permission true if kernels are allowed to read from memory allocated by this pool
-	 * @param write_permission true if kernels are allowed to write to memory allocated by this pool
-	 */
-	void set_access_permissions(const cuda::device_t& device, bool read_permission, bool write_permission)
-	{
-		set_access_permissions(device, access_permissions_t{read_permission, write_permission});
+		return memory::set_permissions(device, *this, permissions);
 	}
 
 	/**
 	 * @param device the devices the kernels running on which are governed by this new setting
-	 * @param permissions new read and write permissions to use
+	 * @param permissions new read and write get_permissions to use
 	 */
 	template <typename DeviceRange>
-	void set_access_permissions(DeviceRange devices, access_permissions_t permissions)
+	void set_permissions(DeviceRange devices, permissions_t permissions)
 	{
-		return memory::set_access_permissions(devices, *this, permissions);
+		return memory::set_permissions(devices, *this, permissions);
 	}
 	///@}
 

src/cuda/api/multi_wrapper_impls/memory.hpp

@@ -512,29 +512,28 @@ inline pool::ipc::imported_ptr_t pool_t::import(const memory::pool::ipc::ptr_han
 	return pool::ipc::import_ptr(*this, exported_handle);
 }
 
-inline access_permissions_t access_permissions(const cuda::device_t& device, const pool_t& pool)
+inline permissions_t get_permissions(const cuda::device_t& device, const pool_t& pool)
 {
-	return cuda::memory::detail_::access_permissions(device.id(), pool.handle());
+	return cuda::memory::detail_::get_permissions(device.id(), pool.handle());
 }
 
-inline void set_access_permissions(const cuda::device_t& device, const pool_t& pool, access_permissions_t permissions)
+inline void set_permissions(const cuda::device_t& device, const pool_t& pool, permissions_t permissions)
 {
 	if (pool.device_id() == device.id()) {
-		throw ::std::invalid_argument("Cannot change the access permissions to a pool of the device "
+		throw ::std::invalid_argument("Cannot change the access get_permissions to a pool of the device "
 			"on which the pool's memory is allocated (" + cuda::device::detail_::identify(device.id()) + ')');
 	}
-	cuda::memory::detail_::set_access_permissions(device.id(), pool.handle(), permissions);
+	cuda::memory::detail_::set_permissions(device.id(), pool.handle(), permissions);
 }
 
 template <typename DeviceRange>
-void set_access_permissions(DeviceRange devices, const pool_t& pool, access_permissions_t permissions)
+void set_permissions(DeviceRange devices, const pool_t& pool, permissions_t permissions)
 {
 	// Not depending on unique_span here :-(
 	auto device_ids = ::std::unique_ptr<cuda::device::id_t[]>(new cuda::device::id_t[devices.size()]);
 	auto device_to_id = [](device_t const& device){ return device.id(); };
 	::std::transform(::std::begin(devices), ::std::end(devices), device_ids.get(), device_to_id);
-	cuda::memory::detail_::set_access_permissions(
-		{ device_ids.get(), devices.size() }, pool.handle(), permissions);
+	cuda::memory::detail_::set_permissions( { device_ids.get(), devices.size() }, pool.handle(), permissions);
 }
 #endif // #if CUDA_VERSION >= 11020