Commit Graph

105 Commits

Author SHA1 Message Date
Daniel Stenberg
897a7b3a13 http2: store upload state per stream
Use a curl_off_t for upload left
2015-05-18 15:41:43 +02:00
Daniel Stenberg
155b1f5df9 http2: fix build when NOT h2-enabled 2015-05-18 14:09:32 +02:00
Daniel Stenberg
979670988a http2: switch to use Curl_hash_destroy()
as after 4883f7019d, the *_clean() function only flushes the hash.
2015-05-18 11:41:16 +02:00
Tatsuhiro Tsujikawa
7ff7e45405 http2: Ignore if we have stream ID not in hash in on_stream_close
We could get stream ID not in the hash in on_stream_close.  For
example, if we decided to reject stream (e.g., PUSH_PROMISE), then we
don't create stream and store it in hash with its stream ID.
2015-05-18 09:33:48 +02:00
Tatsuhiro Tsujikawa
4ac6cc3ebd Require nghttp2 v1.0.0
This commit requires nghttp2 v1.0.0 to compile, and migrate to v1.0.0,
and utilize recent version of nghttp2 to simplify the code,

First we use nghttp2_option_set_no_recv_client_magic function to
detect nghttp2 v1.0.0.  That function only exists since v1.0.0.

Since nghttp2 v0.7.5, nghttp2 ensures header field ordering, and
validates received header field.  If it found error, RST_STREAM with
PROTOCOL_ERROR is issued.  Since we require v1.0.0, we can utilize
this feature to simplify libcurl code.  This commit does this.

Migration from 0.7 series are done based on nghttp2 migration
document.  For libcurl, we removed the code sending first 24 bytes
client magic.  It is now done by nghttp2 library.
on_invalid_frame_recv callback signature changed, and is updated
accordingly.
2015-05-18 09:33:48 +02:00
Daniel Stenberg
077f12b0ae http2: infof length in on_frame_send() 2015-05-18 09:33:48 +02:00
Daniel Stenberg
8b38fcf2f6 http2: bump the h2 buffer size to 32K for speed 2015-05-18 09:33:47 +02:00
Daniel Stenberg
5871affc7a http2: remove the stream from the hash in stream_close callback
... and suddenly things work much better!
2015-05-18 09:33:47 +02:00
Daniel Stenberg
3e8a5d88a5 http2: if there is paused data, do not clear the drain field 2015-05-18 09:33:47 +02:00
Daniel Stenberg
522ca8c035 http2: rename s/data/pausedata 2015-05-18 09:33:47 +02:00
Daniel Stenberg
a66ed407d2 http2: "stream %x" in all outputs to make it easier to search for 2015-05-18 09:33:47 +02:00
Daniel Stenberg
61a94d6f77 http2: Curl_expire() all handles with incoming traffic
... so that they'll get handled next in the multi loop.
2015-05-18 09:33:47 +02:00
Daniel Stenberg
56c362b1a7 http2: don't signal settings change for same values 2015-05-18 09:33:47 +02:00
Daniel Stenberg
72105ebf05 http2: set default concurrency, fix ConnectionExists for multiplex 2015-05-18 09:33:47 +02:00
Daniel Stenberg
38bd6bf0bb bundles: store no/default/pipeline/multiplex
to allow code to act differently on the situation.

Also added some more info message for the connection re-use function to
make it clearer when connections are not re-used.
2015-05-18 09:33:36 +02:00
Daniel Stenberg
be4c8fd1ef http2: lazy init header_recvbuf
It makes us use less memory when not doing HTTP/2 and subsequently also
makes us not have to cleanup HTTP/2 related data when not using HTTP/2!
2015-05-18 08:57:18 +02:00
Daniel Stenberg
783b3c7b42 http2: separate multiplex/pipelining + cleanup memory leaks 2015-05-18 08:57:18 +02:00
Tatsuhiro Tsujikawa
2ce2f03007 http2: Fix bug that data to be drained are overwritten by pending "paused" data 2015-05-18 08:57:18 +02:00
Tatsuhiro Tsujikawa
d722138f29 http2: Don't call nghttp2_session_mem_recv while it is paused by a stream 2015-05-18 08:57:18 +02:00
Tatsuhiro Tsujikawa
0dc0de0351 http2: Read data left in connection buffer after pause
Previously when we do pause because of out of buffer, we just throw
away unread data in connection buffer.  This just broke protocol
framing, and I saw occasional FRAME_SIZE_ERROR.  This commit fix this
issue by remembering how much data read, and in the next iteration, we
process remaining data.
2015-05-18 08:57:18 +02:00
Tatsuhiro Tsujikawa
d261652d42 http2: Fix streams get stuck
This commit fixes the bug that streams get stuck if stream gets some
DATA, and stream->closed becomes true at the same time.  Previously,
in this condition, after we processed DATA, we are going to try to
read data from underlying transport, but there is no data, and gets
EAGAIN.  There was no code path to evaludate stream->closed.
2015-05-18 08:57:18 +02:00
Daniel Stenberg
74a4bd5ecd http2: store incoming h2 SETTINGS 2015-05-18 08:57:18 +02:00
Daniel Stenberg
ee3ad233a2 http2: remove debug logging from on_frame_recv 2015-05-18 08:57:18 +02:00
Daniel Stenberg
70b5b698b0 http2: remove the closed check in http2_recv
With the "drained" functionality we can get here slightly asynchronously
so the stream have have been closed but there is pending data left to
read.
2015-05-18 08:57:18 +02:00
Daniel Stenberg
a28734be09 http2: bump the h2 buffer to 8K 2015-05-18 08:57:18 +02:00
Daniel Stenberg
153f940198 http2: more debug outputs 2015-05-18 08:57:18 +02:00
Daniel Stenberg
f4b8b39881 http2: leave WAITPERFORM when conn is multiplexed
No need to wait for our "spot" like for pipelining
2015-05-18 08:57:18 +02:00
Daniel Stenberg
01e1bdb10c http2: force "drainage" of streams
... which is necessary since the socket won't be readable but there is
data waiting in the buffer.
2015-05-18 08:57:17 +02:00
Daniel Stenberg
7bbac214f5 http2: move the mem+len pair to the stream struct 2015-05-18 08:57:17 +02:00
Daniel Stenberg
84c6b6561f http2: more stream-oriented data, stream ID 0 is for connections 2015-05-18 08:57:17 +02:00
Daniel Stenberg
2c238ea1fc http2: move lots of state data to the 'stream' struct
... from the connection struct. The stream one being the 'struct HTTP'
which is kept in the SessionHandle struct (easy handle).

lookup streams for incoming frames in the stream hash, hashing is based
on the stream id and we get the SessionHandle for the incoming stream
that way.
2015-05-18 08:57:17 +02:00
Daniel Stenberg
5fe71975e4 HTTP: partial start at fixing up hash-lookups on http2 frame receival 2015-05-18 08:57:17 +02:00
Daniel Stenberg
77f1029ecd http: a stream hash for h2 multiplexing 2015-05-18 08:57:17 +02:00
Daniel Stenberg
7957d2eb92 http: a stream hash for h2 multiplexing 2015-05-18 08:57:17 +02:00
Daniel Stenberg
47caff7bdf http2: debug log when receiving unexpected stream_id 2015-05-18 08:54:54 +02:00
Daniel Stenberg
d6440d7366 http2: move stream_id to the HTTP struct (per-stream) 2015-05-18 08:54:54 +02:00
Daniel Stenberg
f858624052 Curl_http2_setup: only do it once and enable multiplex on the server
Once we know we are HTTP/2 enabled we know the server can multiplex.
2015-05-18 08:54:54 +02:00
Daniel Stenberg
54c394699d http2: unify http_conn variable names to 'c' 2015-04-27 22:54:34 +02:00
Tatsuhiro Tsujikawa
21e82bd635 http2: Fix missing nghttp2_session_send call in Curl_http2_switched
Previously in Curl_http2_switched, we called nghttp2_session_mem_recv to
parse incoming data which were already received while curl was handling
upgrade.  But we didn't call nghttp2_session_send, and it led to make
curl not send any response to the received frames.  Most likely, we
received SETTINGS from server at this point, so we missed opportunity to
send SETTINGS + ACK.  This commit adds missing nghttp2_session_send call
in Curl_http2_switched to fix this issue.

Bug: https://github.com/bagder/curl/issues/192
Reported-by: Stefan Eissing
2015-04-02 15:19:29 +02:00
Dan Fandrich
35648f2e79 curl_memory: make curl_memory.h the second-last header file loaded
This header file must be included after all header files except
memdebug.h, as it does similar memory function redefinitions and can be
similarly affected by conflicting definitions in system or dependent
library headers.
2015-03-24 23:47:01 +01:00
Daniel Stenberg
4b02b84897 http2: move the init too for when its actually needed
... it would otherwise lead to memory leakage if we never actually do
the switch.
2015-03-23 10:26:04 +01:00
Daniel Stenberg
abfab1786e HTTP: don't switch to HTTP/2 from 1.1 until we get the 101
We prematurely changed protocol handler to HTTP/2 which made things very
slow (and wrong).

Reported-by: Stefan Eissing
Bug: https://github.com/bagder/curl/issues/169
2015-03-19 13:44:18 +01:00
Daniel Stenberg
0f4a03cbb6 free: instead of Curl_safefree()
Since we just started make use of free(NULL) in order to simplify code,
this change takes it a step further and:

- converts lots of Curl_safefree() calls to good old free()
- makes Curl_safefree() not check the pointer before free()

The (new) rule of thumb is: if you really want a function call that
frees a pointer and then assigns it to NULL, then use Curl_safefree().
But we will prefer just using free() from now on.
2015-03-16 15:01:15 +01:00
Daniel Stenberg
0cf649d9cc http2: detect prematures close without data transfered
... by using the regular Curl_http_done() method which checks for
that. This makes test 1801 fail consistently with error 56 (which seems
fine) to that test is also updated here.

Reported-by: Ben Darnell
Bug: https://github.com/bagder/curl/issues/166
2015-03-14 18:19:51 +01:00
Daniel Stenberg
df28af8f39 http2: make the info-message about receiving HTTP2 headers debug-only 2015-03-07 10:55:37 +01:00
Daniel Stenberg
df5578a7a3 mprintf.h: remove #ifdef CURLDEBUG
... and as a consequence, introduce curl_printf.h with that re-define
magic instead and make all libcurl code use that instead.
2015-03-03 12:36:18 +01:00
Tatsuhiro Tsujikawa
48b5374e65 http2: Return error if stream was closed with other than NO_ERROR
Previously, we just ignored error code passed to
on_stream_close_callback and just return 0 (success) after stream
closure even if stream was reset with error.  This patch records error
code in on_stream_close_callback, and return -1 and use CURLE_HTTP2
error code on abnormal stream closure.
2015-02-27 21:17:27 +00:00
Daniel Stenberg
bc3a44aebc http2: return recv error on unexpected EOF
Pointed-out-by: Tatsuhiro Tsujikawa
Bug: http://curl.haxx.se/bug/view.cgi?id=1487
2015-02-25 13:51:21 +01:00
Daniel Stenberg
b9c190ba77 http2: move lots of verbose output to be debug-only 2015-02-25 11:45:46 +01:00
Tatsuhiro Tsujikawa
7eebf9a3fb http2: Fix bug that associated stream canceled on PUSH_PROMISE
Previously we don't ignore PUSH_PROMISE header fields in on_header
callback.  It makes header values mixed with following HEADERS,
resulting protocol error.
2015-02-09 15:52:56 +01:00