aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2015-05-18pipeline: switch some code over to functionsDaniel Stenberg
... to "compartmentalize" a bit and make it easier to change behavior when multiplexing is used instead of good old pipelining.
2015-05-18symbols-in-versions: add CURLOPT_PIPEWAITDaniel Stenberg
2015-05-18CURLOPT_PIPEWAIT: addedDaniel Stenberg
By setting this option to 1 libcurl will wait for a connection to reveal if it is possible to pipeline/multiplex on before it continues.
2015-05-18Curl_http_readwrite_headers: minor code simplificationDaniel Stenberg
2015-05-18IsPipeliningPossible: fixed for http2Daniel Stenberg
2015-05-18http2: bump the h2 buffer size to 32K for speedDaniel Stenberg
2015-05-18http2: remove the stream from the hash in stream_close callbackDaniel Stenberg
... and suddenly things work much better!
2015-05-18http2: if there is paused data, do not clear the drain fieldDaniel Stenberg
2015-05-18http2: rename s/data/pausedataDaniel Stenberg
2015-05-18http2: "stream %x" in all outputs to make it easier to search forDaniel Stenberg
2015-05-18http2: Curl_expire() all handles with incoming trafficDaniel Stenberg
... so that they'll get handled next in the multi loop.
2015-05-18http2: don't signal settings change for same valuesDaniel Stenberg
2015-05-18http2: set default concurrency, fix ConnectionExists for multiplexDaniel Stenberg
2015-05-18bundles: store no/default/pipeline/multiplexDaniel Stenberg
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.
2015-05-18http2: lazy init header_recvbufDaniel Stenberg
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!
2015-05-18http2: separate multiplex/pipelining + cleanup memory leaksDaniel Stenberg
2015-05-18CURLMOPT_PIPELINE: bit 1 is for multiplexingDaniel Stenberg
2015-05-18http2: Fix bug that data to be drained are overwritten by pending "paused" dataTatsuhiro Tsujikawa
2015-05-18http2: Don't call nghttp2_session_mem_recv while it is paused by a streamTatsuhiro Tsujikawa
2015-05-18http2: Read data left in connection buffer after pauseTatsuhiro Tsujikawa
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.
2015-05-18http2: Fix streams get stuckTatsuhiro Tsujikawa
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.
2015-05-18http2: store incoming h2 SETTINGSDaniel Stenberg
2015-05-18pipeline: move function to pipeline.c and make staticDaniel Stenberg
... as it was only used from there.
2015-05-18IsPipeliningPossible: http2 can always "pipeline" (multiplex)Daniel Stenberg
2015-05-18http2: remove debug logging from on_frame_recvDaniel Stenberg
2015-05-18http2: remove the closed check in http2_recvDaniel Stenberg
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.
2015-05-18http2: bump the h2 buffer to 8KDaniel Stenberg
2015-05-18http2: Curl_read should not use the single bufferDaniel Stenberg
... as it does for pipelining when we're multiplexing, as we need the different buffers to store incoming data correctly for all streams.
2015-05-18http2: more debug outputsDaniel Stenberg
2015-05-18http2: leave WAITPERFORM when conn is multiplexedDaniel Stenberg
No need to wait for our "spot" like for pipelining
2015-05-18http2: force "drainage" of streamsDaniel Stenberg
... which is necessary since the socket won't be readable but there is data waiting in the buffer.
2015-05-18http2: move the mem+len pair to the stream structDaniel Stenberg
2015-05-18http2: more stream-oriented data, stream ID 0 is for connectionsDaniel Stenberg
2015-05-18http2: move lots of state data to the 'stream' structDaniel Stenberg
... 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.
2015-05-18HTTP: partial start at fixing up hash-lookups on http2 frame receivalDaniel Stenberg
2015-05-18http: a stream hash for h2 multiplexingDaniel Stenberg
2015-05-18http: a stream hash for h2 multiplexingDaniel Stenberg
2015-05-18http2: debug log when receiving unexpected stream_idDaniel Stenberg
2015-05-18http2: move stream_id to the HTTP struct (per-stream)Daniel Stenberg
2015-05-18Curl_http2_setup: only do it once and enable multiplex on the serverDaniel Stenberg
Once we know we are HTTP/2 enabled we know the server can multiplex.
2015-05-18http: switch on "pipelining" (multiplexing) for HTTP/2 serversDaniel Stenberg
... and do not blacklist any.
2015-05-15README.pipelining: removedDaniel Stenberg
All the details mentioned here are better documented in man pages
2015-05-14build: removed bundles.c from make filesDan Fandrich
This file was removed in commit fd137786
2015-05-14Curl_conncache_add_conn: fix memory leak on OOMDaniel Stenberg
2015-05-12CURLMOPT_MAX_HOST_CONNECTIONS: host = host name + port numberDaniel Stenberg
2015-05-12conncache: keep bundles on host+port bases, not only host namesDaniel Stenberg
Previously we counted all connections to a specific host name and that would be used for the CURLMOPT_MAX_HOST_CONNECTIONS check for example, while servers on different port numbers are normally considered different "origins" on the web and should thus be considered different hosts.
2015-05-12bundles: merged into conncache.cDaniel Stenberg
All the existing Curl_bundle* functions were only ever used from within the conncache.c file, so I moved them over and made them static (and removed the Curl_ prefix).
2015-05-12hostcache: made all host caches use structs, not pointersDaniel Stenberg
This avoids unnecessary dynamic allocs and as this also removed the last users of *hash_alloc() and *hash_destroy(), those two functions are now removed.
2015-05-12multi: converted socket hash into non-allocated structDaniel Stenberg
avoids extra dynamic allocation
2015-05-12connection cache: avoid Curl_hash_alloc()Daniel Stenberg
... by using plain structs instead of pointers for the connection cache, we can avoid several dynamic allocations that weren't necessary.