Commit graph

169 commits

Author SHA1 Message Date
918f8241be Fix invalid use of mutex in TSLQueue
Mutex was removed in favor of the custom SharedSpinLock.
2024-01-11 21:05:35 +09:00
4c8cf82a31 Don't fail on "try" fns if failed to get spinLock
In SharedSpinLock: Only fail on "try" fns after spinLock was acquired
and condition is not met.
2024-01-11 21:05:35 +09:00
fe1eefdd7c Refactor shared-spin-lock to use atomic "spinLock" 2024-01-11 21:05:35 +09:00
3212f3e7dc Minor refactorings 2024-01-11 21:05:35 +09:00
abd65b9a94 Reland C++11 "shared_lock" with iter remove fix
On iterator remove, the iterator will trade the read lock for a write
lock, and trade back for a read lock once the remove has been completed.
2024-01-11 21:05:35 +09:00
c03eae1c15 UnitTest for start/stop threaded update and fix
All checks were successful
Publish doxygen documentation to seodisparate.com / doxygen-gen-and-publish (push) Successful in 0s
2024-01-11 20:07:25 +09:00
de2848004f Enforce thread-safety on destroy context 2024-01-11 19:40:15 +09:00
186f2edf0f Use mutex when enabling/disabling threaded-update 2024-01-11 19:00:27 +09:00
74341e83d4 Use atomic_bool for auto-updating "flag" 2024-01-11 18:45:14 +09:00
5d24c4b5e6 Add note about usage of UDPC header in docs
All checks were successful
Publish doxygen documentation to seodisparate.com / doxygen-gen-and-publish (push) Successful in 0s
2024-01-01 17:15:55 +09:00
f4ed1aa136 Note in documentation about thread safety
All checks were successful
Publish doxygen documentation to seodisparate.com / doxygen-gen-and-publish (push) Successful in 0s
Basically, all functions should be thread-safe unless mentioned
otherwise.
2024-01-01 12:25:04 +09:00
611287b377 Revert "Impl "RWLock" for use in TSLQueue"
This reverts commit cf27a6bb76.

The use of "shared_lock" in TSLQueue is unsafe because of two things:

  - The TSLQueue iterator takes a "read" lock.
  - The TSLQueue iterator can erase the current element.
2023-07-22 17:28:33 +09:00
ce092f81f9 Revert "Replace unnecessary use of atomic in "shared_lock""
This reverts commit b50e50b5fb.
2023-07-22 17:28:13 +09:00
b50e50b5fb Replace unnecessary use of atomic in "shared_lock" 2023-07-22 16:46:38 +09:00
cf27a6bb76 Impl "RWLock" for use in TSLQueue
This project supports C++11, and std::shared_lock was made available in
C++17, thus a "shared_spin_lock" was created with similar functionality.
This "shared_spin_lock" is used in TSLQueue.
2023-07-22 16:33:09 +09:00
64a0995e21 Fix CMakeLists.txt, handle -Weffc++ warnings 2023-07-22 13:25:43 +09:00
97f030a3a3 Add docs for UDPC_no* functions 2023-07-04 11:41:45 +09:00
972ec73e88 Refactor addr logprint via template specialization 2023-06-26 14:01:42 +09:00
77d69cabbc Impl "unsafe" versions of UDPC_atostr(...)
These "unsafe" versions are guaranteed to not have the returned address
strings be overwritten by UDPC, but they must be manually free'd later
(as mentioned in the documentation).
2023-06-22 13:25:26 +09:00
4abd9fddcb Conditionally use UDPC_atostr internally
Previous implementation used UDPC_atostr() frequently. This commit
changes the implementation to only call UDPC_atostr() if the log level
of the code logging to output will cause the log to be output.
2023-06-22 12:39:11 +09:00
390f3972f8 Add documentation for UDPC_atostr... 2023-06-22 12:07:26 +09:00
2495396d76 Fix potential memory corruption bug
UDPC_atostr(...) uses a uint32_t as an offset into a buffer inside the
UDPC Context such that there can be at most 32 different addr-strings.
The call is thread-safe, up to a point (at most 32 concurrent calls will
return correct strings). The problem was that the offset was not being
reset to 0 and was always being incremented by 40. Should the offset
overflow, the offset into the buffer will be "mis-aligned" such that the
32nd offset into the buffer can possibly overwrite memory past the
buffer's end if the addr string is long enough.

The fix is to use a mutex instead of an atomic uint32_t to lock a
code-block that will always prevent overflow (using modulus operator).

I think the speed-loss is negligable (using a lock instead of an atomic
uint32_t) since it isn't expected for programs using UDPC to use
UDPC_atostr(...) very frequently.
2023-06-21 13:23:26 +09:00
211715fc56 Fix potential nullptr deref in free_PacketInfo_ptr 2023-04-19 19:11:03 +09:00
3fac706d86 Add UDPC_free_PacketInfo_ptr(...)
Also add unit test for this function.
2023-04-19 18:50:26 +09:00
f18e87903a Refactor UDPC::get_empty_pinfo() 2023-04-18 14:03:31 +09:00
087d3cfc6f Impl handling Ctrl-C for Windows in NetworkTest 2023-01-10 12:14:57 +09:00
577ab0bc52 Better handling of receiving packets in recv loop
On error processing a received packet, continue to receive packets in
the loop. If no packet was received, the loop ends.
2023-01-10 11:51:54 +09:00
a8ef89d9b1 Update documentation about UDPC_update(ctx) 2022-12-08 16:54:02 +09:00
77ac7f88a5 Receive packets in loop til none left in interval
Also some formatting fixes (max 80 chars in edited code).
2022-12-04 19:58:13 +09:00
e774a54349 Handle SIGINT (Ctrl-C) in NetworkTest 2022-11-23 14:13:44 +09:00
e08e620b26 Add |id| to PacketInfo 2021-01-10 14:40:17 +09:00
ba3d6df406 Use dll_export (for builds for Windows) 2020-06-23 13:01:27 +09:00
b0e996cda3 Attempt to make NetworkTest work on Windows 2020-04-28 20:55:16 +09:00
2a39267b38 Add/fix more documentation 2020-04-28 19:38:07 +09:00
d3fb621406 Minor fixes 2020-04-27 19:34:04 +09:00
445a221a5b Fix check addr endianness in a4toa6 function 2020-04-27 11:47:50 +09:00
31ca4ddc44 Impl create id with hostname
Previously, ids could only be created with an ip address. Now they can
be made with a hostname, which will be looked up by UDPC.

Also fix client still requesting connections even if
accept-new-connections flag is false.
2020-04-26 18:52:06 +09:00
5dcab1d590 Minor fix to documentation 2020-04-18 16:19:59 +09:00
76c5bb750d Use memcpy instead of pointer casted assignment
std::memcpy is used over int pointer casted assignment due to possible
alignment issues that may cause the casted pointer to be invalid.
2020-04-18 16:09:00 +09:00
cf6ff5a040 Change how UDPC_PacketInfo handles it's data
The "data" member variable in UDPC_PacketInfo is now handled as a
pointer to dynamic data, instead of an array with a fixed size. Every
time a UDPC_PacketInfo is received from the context,
UDPC_free_PacketInfo() must be called on it to avoid a memory leak.
2020-04-15 19:56:15 +09:00
e01a1ccd94 Fix compilation fail on no libsodium 2020-03-06 13:03:47 +09:00
75f54119bf Fix use of C++14/C++17, only C++11 is supported 2020-03-06 12:05:33 +09:00
c37909bde3 Add helpers converting network-order (big-endian) 2020-01-16 20:28:42 +09:00
0170423a35 Fix UDPC::isBigEndian returning always true 2020-01-16 20:28:01 +09:00
136c8b21a5 Fix verification message when using libsodium
Previous implementation had the client send only epoch-time-in-seconds
to be signed by the server. Now the client sends random data and
epoch-time to be signed by the server.
2020-01-15 16:31:38 +09:00
83392fc916 Fix doxygen documentation 2020-01-13 19:22:08 +09:00
b41639c568 Set pointers to const where possible in API 2020-01-10 20:28:08 +09:00
f588d409c9 Truncate "UDPConnection" to "UDPC" 2020-01-09 16:57:01 +09:00
40c4c9b2ba Fix potential non-null-terminated string bug 2020-01-09 16:45:57 +09:00
4d68715efb Add more documentation, some fixes 2020-01-08 20:50:56 +09:00