PS-10068 Production code for new KMIP C++ library #23

lukin-oleksiy · 2025-12-22T10:57:26Z

The "kmipclient" C++ library implementation is added, that wraps low-level "kmip.h" calls and bypasses mid-level "kmip_bio" level. The kmip.h file was splitted in 3 to export enums, structures and functions separately.
Comparing to the 1st version (POC), The interface was simplified and uses exceptions now.
There are plans to replace C implementation of the protocol serialization/deserialization (kmip.c and others) with C++ code, so the current version of "kmipclient" library contains a lot of TODO and skeletons for future work. Ath this stage of development, the middle-level C implementation (kmip_bio.c) is removed and the library works directly with kmip.c.
Previous PR of POC is closed.

The "kmipclient" C++ library implementation is added, that wraps low-level "kmip.h" calls and bypasses mid-level "kmip_bio" level.

https://perconadev.atlassian.net/browse/PS-10068 Addressing PR comments and other fixes, finallizing the interface of the library, adding test suite

percona-ysorokin · 2026-01-13T12:27:00Z

kmipclient/CMakeLists.txt

+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address -g")
+set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=address -g")
+
+include_directories(${PROJECT_SOURCE_DIR}/include)


Change to target_include_directories

percona-ysorokin · 2026-01-13T12:30:20Z

kmipclient/CMakeLists.txt

+set(CMAKE_CXX_STANDARD 23)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF) # Optional, but recommended for standard compliance


Use target-specific options

target_compile_features(target INTERFACE cxx_std_20)

set_target_properties(my_target PROPERTIES CXX_STANDARD_REQUIRED YES CXX_EXTENSIONS NO )

Fixed. Kid of. I still need 2 statements at the top of CMakeLists.txt to compile

percona-ysorokin · 2026-01-13T12:34:37Z

kmipclient/CMakeLists.txt

+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address -g")
+set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=address -g")


Unconditionaly always enabled Address Sanitazer looks weird to me. I would declare an option

option(WITH_ASAN "Enable Address Sanitizer" OFF) if(WITH_ASAN) target_compile_options(mytarget INTERFACE "-fsanitize=address") target_link_options(mytarget INTERFACE "-fsanitize=address") endif()

percona-ysorokin · 2026-01-13T12:34:56Z

kmipclient/CMakeLists.txt

+)
+
+target_link_libraries(kmipclient kmip)
+set_property(TARGET kmipclient PROPERTY POSITION_INDEPENDENT_CODE ON)


set_target_property()

Not fixed, there's no such funtion

percona-ysorokin · 2026-01-13T12:36:43Z

kmipclient/CMakeLists.txt

+
+macro(add_example name)
+  add_executable(example_${name} examples/example_${name}.cpp)
+  target_link_libraries(example_${name} kmipclient)


Suggested change

target_link_libraries(example_${name} kmipclient)

target_link_libraries(example_${name} PRIVATE kmipclient)

percona-ysorokin · 2026-01-13T13:27:09Z

kmipclient/src/KmipClient.cpp

+KmipClient::KmipClient (NetClient &net_client, const std::shared_ptr<Logger> &log) : net_client (net_client)
+{
+  io     = std::make_unique<IOUtils> (net_client);
+  logger = log;
+}


Suggested change

KmipClient::KmipClient (NetClient &net_client, const std::shared_ptr<Logger> &log) : net_client (net_client)

{

io = std::make_unique<IOUtils> (net_client);

logger = log;

}

KmipClient::KmipClient (NetClient &net_client, const std::shared_ptr<Logger> &log) : net_client (net_client),

io(std::make_unique<IOUtils> (net_client)),

logger(log)

{}

Fixed in all constructors

percona-ysorokin · 2026-01-13T13:28:30Z

kmipclient/src/KmipCtx.hpp

+  int       buffer_blocks     = 1;
+  const int buffer_block_size = 1024;


Again, signed types for sizes

Well, the reason is signed types in the lower level functions (kmip.h).

percona-ysorokin · 2026-01-13T13:30:08Z

kmipclient/src/KmipRequest.hpp

+inline void
+KmipRequest::add_batch_item (RequestBatchItem *rbi)
+{
+  // Sorry, C++ guys, we have to use address arithmetic here because of the lower level


Yeah, I know, I know... The goal of the next development iteration is to replace kmip.c with KmipSerializer and KmipDeserializer classes and get rid of all malloc's and address arithmetic. So, ... let's keep it for a while this way. You can't imageine what a mess we have in a "heritage" code.

percona-ysorokin · 2026-01-13T13:30:41Z

kmipclient/src/NetClientOpenSSL.cpp

+  return BIO_read (bio_, data, dlen);
+}
+
+} // namespace


No EOL at the end of file

percona-ysorokin · 2026-01-13T13:31:39Z

kmipclient/src/RequestFactory.hpp

+  static void create_get_rq (KmipCtx &ctx, const id_t &id);
+  static void create_activate_rq (KmipCtx &ctx, const id_t &id);
+  static void create_create_aes_rq (KmipCtx &ctx, const name_t &name, const name_t &group);
+  static void create_register_key_rq (KmipCtx &ctx, const name_t &name, const name_t &group, const Key &k);
+  static void create_register_secret_rq (KmipCtx &ctx, const name_t &name, const name_t &group, std::string &secret,
+                                         int secret_data_type);
+  static void create_revoke_rq (KmipCtx &ctx, const id_t &id, int reason, const name_t &message,
+                                time_t occurrence_time);
+  static void create_destroy_rq (KmipCtx &ctx, const id_t &id);
+  static void create_get_attributes_rq (KmipCtx &ctx, const id_t &id, const std::vector<std::string> &attr_names);
+  static void create_get_attribute_list_rq (KmipCtx &ctx, const id_t &id);
+  static void create_locate_by_name_rq (KmipCtx &ctx, const name_t &name, enum object_type o_type, int max_items,
+                                        int offset);
+  static void create_locate_by_group_rq (KmipCtx &ctx, const name_t &group_name, enum object_type o_type, int max_items,
+                                         size_t offset);
+  static void create_locate_all_rq (KmipCtx &ctxm, enum object_type o_type, int max_items, int offset);
+
+private:
+  static void create_locate_rq (KmipCtx &ctx, bool is_group, const name_t &name, enum object_type o_type, int max_items,
+                                size_t offset);


remove enum for types, use std::size_t for sizes

PS-9697 POC prototype of KMIP client library

fbcabbd

The "kmipclient" C++ library implementation is added, that wraps low-level "kmip.h" calls and bypasses mid-level "kmip_bio" level.

lukin-oleksiy requested review from dutow, ktrushin and percona-ysorokin December 22, 2025 10:57

PS-10068 Making KMIP C++ library productgion ready

6da142e

https://perconadev.atlassian.net/browse/PS-10068 Addressing PR comments and other fixes, finallizing the interface of the library, adding test suite

lukin-oleksiy force-pushed the PS-9697-POC-of-kmip-cpp branch from f5f8fbc to 6da142e Compare January 8, 2026 10:37

percona-ysorokin reviewed Jan 13, 2026

View reviewed changes

Fixing PR comments

0aea004

lukin-oleksiy force-pushed the PS-9697-POC-of-kmip-cpp branch from c407d1d to 95a4c5f Compare January 21, 2026 11:09

Fixes of include directories

8e534f5

lukin-oleksiy force-pushed the PS-9697-POC-of-kmip-cpp branch from 95a4c5f to 8e534f5 Compare January 21, 2026 11:10

Update of the interface

c86e1e8

		set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address -g")
		set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=address -g")

	target_link_libraries(example_${name} kmipclient)
	target_link_libraries(example_${name} PRIVATE kmipclient)

PS-10068 Production code for new KMIP C++ library #23

Are you sure you want to change the base?

PS-10068 Production code for new KMIP C++ library #23

Uh oh!

Conversation

lukin-oleksiy commented Dec 22, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants