QUIC API: More minor tweaks

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19769)

diff --git a/doc/designs/quic-design/quic-api-ssl-funcs.md b/doc/designs/quic-design/quic-api-ssl-funcs.md
index 1a912dc35b54159bcdfa348c4175a71c3952fb79..d43e58684aebd07123dce53b2c63d68021ab7821 100644 (file)
--- a/doc/designs/quic-design/quic-api-ssl-funcs.md
+++ b/doc/designs/quic-design/quic-api-ssl-funcs.md
@@ -622,7 +622,6 @@ Notes:
 | `SSL_tick` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
 | `SSL_get_tick_timeout` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
 | `SSL_get_blocking_mode` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
-| `SSL_get_blocking_mode` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
 | `SSL_set_blocking_mode` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
 | `SSL_get_rpoll_descriptor` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |
 | `SSL_get_wpoll_descriptor` | CSSM | 🟦N | 🟩A | 🟥QSA | 🟢Done |

diff --git a/doc/designs/quic-design/quic-api.md b/doc/designs/quic-design/quic-api.md
index 2f3f151f93e0a7d87d1c00859bb60e68e886284b..bb7bc28154282791f63831e99aa7cf5c115426b8 100644 (file)
--- a/doc/designs/quic-design/quic-api.md
+++ b/doc/designs/quic-design/quic-api.md
@@ -394,7 +394,8 @@ Should not require any changes.
 
 - `SSL_MODE_ENABLE_PARTIAL_WRITE`: Implemented. If this mode is set during a
   non-partial-write `SSL_write` operation spanning multiple `SSL_write` calls,
-  this operation is aborted and partial write mode begins immediately.
+  this mode does not take effect until the non-partial write operation is
+  completed.
 
 - `SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER`: Implemented.
 
@@ -424,9 +425,6 @@ Advances the QUIC state machine to the extent feasible, potentially performing
 network I/O. Also compatible with DTLSv1 and supercedes `DTLSv1_handle_timeout`
 for all use cases.
 
-TBD: Should we just map this to DTLS_CTRL_HANDLE_TIMEOUT internally (and maybe
-alias the CTRL #define)?
-
 TBD: Deprecate `DTLSv1_get_timeout`?
 TBD: Deprecate `DTLSv1_handle_timeout`?
 
@@ -1185,3 +1183,12 @@ Where connection closure is initiated remotely rather than locally, only the
 draining state is relevant. Since we conclude above that we do not need to
 implement the draining state on the client side, this means that connection
 closure can be completed immediately in the case of a remote closure.
+
+**Q. Should we just map `SSL_tick` to `DTLS_CTRL_HANDLE_TIMEOUT` internally?**
+
+A. No, since the infinite time representation is different between the two
+calls.
+
+**Q. How should `STOP_SENDING` be supported?**
+
+TODO: Determine how `STOP_SENDING` should be supported.