Fix the bug that causes a stream to stall when a receiver, which
enables nghttp2_option_set_no_auto_window_update() option on, sends
SETTINGS_INITIAL_WINDOW_SIZE with the value that is less than or equal
to the amount of data received. Previously, in this particular case,
when SETTINGS is acknowledged by the sender, the receiver does not try
to send WINDOW_UPDATE frame. The sender is unable to send more data
because its stream-level window size is smaller than or equal to the
amount of data it has sent.
When applying new header table size acknowledged with SETTINGS ACK by
an encoder, change the header table size on a decoder only when it
strictly lowers the current maximum table size set by Dynamic Table
Size Update from the encoder or the default size 4096 if no Dynamic
Table Size Update is received.
Previously, the header table size on a decoder is always changed. If
a maximum size in SETTINGS are increased (e.g., 4096 -> 8192), and
then decreased to the previous value, the decoder incorrectly requires
Dynamic Table Size Update from an encoder.
Previously, if automatic window update is enabled (which is default),
after window size is set to 0 by
nghttp2_session_set_local_window_size, once the receiving window is
exhausted, even after window size is increased by
nghttp2_session_set_local_window_size, no more data cannot be
received. This is because nghttp2_session_set_local_window_size does
not submit WINDOW_UPDATE. It is only triggered when new data arrives
but since window is filled up, no more data cannot be received, thus
dead lock happens.
This commit fixes this issue. nghttp2_session_set_local_window_size
submits WINDOW_UPDATE if necessary.
https://github.com/curl/curl/issues/4939
Add the currently-unused `test_nghttp2_session_create_idle_stream()`
function to the test suite definition.
Modify the test in two places to make it pass:
* Use stream ID=10 as the priority stream ID to test automatic creation
of streams for priority specs. The code below checks against stream
ID=10 so I assume this was a typo in the test.
* Set the `last_sent_stream_id` instead of the `next_stream_id` to test
that idle streams cannot be created with smaller numbers than the
most-recently-seen stream ID. Looking at the validation path in
`session_detect_idle_stream()`, I think this was another test typo.
This commit fixes the bug that stream is closed with wrong error code
(0). This happens when STREAM or DATA frame with END_STREAM flag set
is received and it violates HTTP messaging rule (i.e., content-length
does not match) and the other side of stream has been closed. In this
case, nghttp2_on_stream_close_callback should be called with nonzero
error code, but previously it is called with 0 (NO_ERROR).
The maximum number of outgoing concurrent streams is initially
limited to 100 to avoid issues when the local endpoint submits
lots of requests before receiving initial SETTINGS frame from
the remote endpoint, since sending them at once to the remote
endpoint could lead to rejection of some of the requests.
This initial limit is overwritten with the value advertised in
SETTINGS_MAX_CONCURRENT_STREAMS setting by the remote endpoint,
but previously, it wasn't lifted if the remote endpoint didn't
advertise that setting (implying no limits), in which case the
limit of 100 was retained, even though it was never advertised
by the remote endpoint.
Signed-off-by: Piotr Sikora <piotrsikora@google.com>
Previously, the incoming invalid regular header field was ignored by
default. With this commit, they are now treated as stream error, and
the stream is reset by default. The error code used is now
PROTOCOL_ERROR, instead of INTERNAL_ERROR.
nghttp2_option_no_closed_streams controls whether closed streams are
retained or not. If nonzero is passed to that function's parameter
val, a session does not retain closed streams. It may hurt the shape
of priority tree, but can save memory.
Revert part of 16c46114dc to fix race
condition that incoming stream after sending GOAWAY causes connection
error. The strict stream handling introduced in the above commit does
not handle several cases well (e.g., GOAWAY race, and refusing streams
because of concurrency limit).
Tatsuhiro Tsujikawa
Dimitris Apostolou
James M Snell
Tatsuhiro Tsujikawa
Geoff Hill
Piotr Sikora