- Jan 21, 2020
-
-
Tatsuhiro Tsujikawa authored
-
- Jan 18, 2020
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
- Jan 14, 2020
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
- Jan 12, 2020
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
James M Snell authored
* `ngtcp2_crypto_generate_stateless_reset_token` - Used to generate a stateless reset token as an HKDF extraction using the CID and a token secret as input.
-
Tatsuhiro Tsujikawa authored
-
- Jan 11, 2020
-
-
Tatsuhiro Tsujikawa authored
Remove unused color-related code from examples
-
- Jan 09, 2020
-
-
Amir Livneh authored
-
Tatsuhiro Tsujikawa authored
-
- Jan 05, 2020
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
- Jan 04, 2020
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
- Dec 28, 2019
-
-
Tatsuhiro Tsujikawa authored
-
- Dec 21, 2019
-
-
Tatsuhiro Tsujikawa authored
Fix typo
-
- Dec 19, 2019
-
-
Amir Livneh authored
-
- Dec 18, 2019
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
Fix example server's write callback never being called
-
- Dec 12, 2019
-
-
Amir Livneh authored
Handler::wev_ is initialized with a file descriptor set to 0. ev_io_set() is never called for this watcher, so its file descriptor remains 0 and calls to ev_io_start() don't have the desired effect. As a result writecb() is never called. There may still be an issue due to all endpoints sharing the same watcher, but this is mostly orthogonal to the problem fixed here.
-
Tatsuhiro Tsujikawa authored
Avoid use-after-free if key update is initiated while ppe is pending
-
- Dec 09, 2019
-
-
Amir Livneh authored
conn_rotate_keys() frees conn->pktns->crypto.tx.ckm. If there is a pending ppe, the ppe object still holds a reference to the freed memory and will eventually try to use it in ngtcp2_ppe_final().
-
- Dec 08, 2019
-
-
Tatsuhiro Tsujikawa authored
Fix use of uninitialized data in tests
-
- Dec 01, 2019
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-
- Nov 30, 2019
-
-
Amir Livneh authored
write_single_frame_pkt() returns a packet length greater than the number of bytes it writes to the buffer. The returned length reflects a fake AEAD overhead, while this extra overhead is never written. As a result, ngtcp2_verify_stateless_reset_token() uses initialized data when reading the end of the packet. This causes noise when using a memory sanitizer.
-
- Nov 27, 2019
-
-
Tatsuhiro Tsujikawa authored
-
- Nov 22, 2019
-
-
Tatsuhiro Tsujikawa authored
-
Tatsuhiro Tsujikawa authored
-