Loïc Hoguin
28d3515d71
Remove a bunch of todos
2018-04-30 18:39:27 +02:00
Loïc Hoguin
8d1f468ac0
Reject HTTP/2 requests with a body size different than content-length
2018-04-30 13:47:33 +02:00
Loïc Hoguin
84b4128d06
Receive and ignore HTTP/2 request trailers if any
...
This is a first step toward properly supporting request trailers.
2018-04-29 17:39:48 +02:00
Loïc Hoguin
2db5ffbf84
Add SETTINGS ack timeout and option settings_timeout
2018-04-28 11:00:50 +02:00
Loïc Hoguin
3d6bb01d5f
Reject WINDOW_UPDATE frames sent after an RST_STREAM
2018-04-27 20:45:34 +02:00
Loïc Hoguin
9a29aea148
Add options controlling maximum h2 frame sizes
2018-04-27 17:58:37 +02:00
Loïc Hoguin
d38d86c4a9
Add options controlling initial control flow windows
2018-04-26 22:08:05 +02:00
Loïc Hoguin
7373822b86
Add the max_concurrent_streams h2 option
2018-04-25 21:32:58 +02:00
Loïc Hoguin
8f4adf437c
Add options to control h2's SETTINGS_HEADER_TABLE_SIZE
2018-04-25 16:55:52 +02:00
Loïc Hoguin
8b9a09c9fe
HTTP/2 informational responses don't end the stream
2018-04-23 15:49:34 +02:00
Loïc Hoguin
9af8b0d23a
Add missing enable_connect_protocol in typespecs
2018-04-23 14:34:53 +02:00
Loïc Hoguin
adf1a40287
Fix Dialyzer warnings
2018-04-04 17:34:10 +02:00
Loïc Hoguin
bbfc1569cc
Add initial implementation of Websocket over HTTP/2
...
Using the current draft:
https://tools.ietf.org/html/draft-ietf-httpbis-h2-websockets-01
2018-04-04 17:23:37 +02:00
Kirill Kinduk
4c34774b7e
Add max_frame_size option for websocket handlers
...
Option allows to limit a frame by size before decoding its payload.
LH: I have added a test for when the limit is reached on a nofin
fragmented frame (the last commit addressed that case but it had
no test). I have fixed formatting and other, and changed the
default value to infinity since it might otherwise be incompatible
with existing code. I also added documentation and a bunch of other
minor changes.
2018-03-28 16:58:02 +02:00
Loïc Hoguin
5964273cc4
Fix the flushing of messages when switching to Websocket
...
We now flush messages that are specific to cowboy_http only.
Stream handlers should also flush their own specific messages
if necessary, although timeouts will be flushed regardless
of where they originate from.
Also renames the http_SUITE to old_http_SUITE to distinguish
new tests from old tests. Most old tests need to be removed
or converted eventually as they're legacy tests from Cowboy 1.0.
2018-03-26 18:53:07 +02:00
Loïc Hoguin
4d5174632c
Add initial support for system messages in cowboy_loop
2018-03-23 18:33:16 +01:00
Loïc Hoguin
21c9c66971
Merge the two separate receive loops in cowboy_websocket
...
Also rename a bunch of functions to make the code easier to read.
2018-03-23 16:32:53 +01:00
Loïc Hoguin
6945882cd3
Fix a Dialyzer warning
2018-03-14 17:43:13 +01:00
Bartek Walkowicz
2131709328
Add case for handling infinity for idle/request_timeout
...
Currently cowboy assumes that idle_timeout or request_timeout is
a number and always starts timers. Similar situation takes place
in case of preface_timeout for http2. This commit adds case for
handling infinity as a timeout, allowing to not start mentioned
timers.
2018-03-14 17:15:06 +01:00
Loïc Hoguin
f9092126fa
Handle system messages in cowboy_websocket
2018-03-13 11:19:13 +01:00
Loïc Hoguin
b9c8d86502
Handle supervisor calls properly everywhere
2018-03-13 10:40:14 +01:00
Loïc Hoguin
19054e40e0
Fix crash in cowboy_http2 when content-length is invalid
2018-02-28 16:18:29 +01:00
Loïc Hoguin
819a47a220
Fix sending HTTP/2 bodies as iolists out of order
2018-01-24 11:44:03 +01:00
Loïc Hoguin
e09ae60d5b
Fix an unused variable warning
2018-01-23 16:29:25 +01:00
Loïc Hoguin
482de55a96
Fix a miscount of output flow control window for HTTP/2
...
The miscount occurred because of a faulty iolist split function.
The bug should now be corrected, a PropEr test has been added
and a regression test has also been added.
2018-01-23 16:14:18 +01:00
Loïc Hoguin
3a7b411143
Don't crash in delete_resp_header if no resp headers are set
2018-01-22 14:00:05 +01:00
Paul Fisher
82cd22a88b
Allow colon within path segments
...
Allow `cowboy_router:compile` to handle colon characters within
path segments, rather than exiting with `badarg`. This is allowed
via RFC 7230 2.7 -> [RFC 3986 3.3](https://tools.ietf.org/html/rfc3986#section-3.3 ):
```
segment = *pchar
segment-nz = 1*pchar
segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" )
; non-zero-length segment without any colon ":"
pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
```
2018-01-16 11:24:49 +01:00
Loïc Hoguin
34473bc247
Fix Dialyzer warnings
2017-12-13 12:40:00 +01:00
Loïc Hoguin
576e2d2111
Don't badmatch on HTTP/2 preface's Transport:send
...
Send errors produce annoying logs and we notice the connection
is gone later on anyway.
2017-12-12 16:57:05 +01:00
Loïc Hoguin
eafca21198
Fix HTTP/1.1 pipelining
...
Cases where a request body was involved could sometimes
fail depending on timing. Also fix all of the old
http_SUITE tests.
2017-12-12 11:37:48 +01:00
Loïc Hoguin
17719a136d
Add the few remaining tests to the rfc7231 test suite
2017-12-08 20:38:31 +01:00
Loïc Hoguin
b000d53855
Add more rfc7231 tests and a new max_skip_body_length option
...
The option controls how much body we accept to skip for HTTP/1.1
connections when the user code did not consume the body fully.
It defaults to 1MB.
2017-12-07 22:33:52 +01:00
Loïc Hoguin
c2b813684e
Send a 426 when Websocket is required and client didn't upgrade
2017-12-06 17:31:32 +01:00
Loïc Hoguin
5269bf580b
Fix 408 not sending connection: close for HTTP/1.1
...
Also make sure the header is sent for all types of early_error
that result in the closing of the connection.
2017-12-06 16:30:30 +01:00
Loïc Hoguin
bc39b433bb
Properly handle OPTIONS * requests
...
Support for these was broken during the development
of Cowboy 2.0. It is now fixed and better handled
than it ever was.
2017-12-06 14:05:30 +01:00
Loïc Hoguin
2eb3e3f994
Also disable the TRACE method entirely
2017-12-06 10:54:23 +01:00
Loïc Hoguin
dd002b8141
Disable the CONNECT method completely
...
It's safer than allow it with the wrong behavior.
2017-12-06 00:30:59 +01:00
Loïc Hoguin
10dc2c2ef0
Add an rfc7231 test suite, fix an HTTP/2 bug with HEAD
...
In some cases there could be a body sent as a response to
a HEAD request when using HTTP/2. This has been corrected.
2017-12-04 18:21:10 +01:00
Loïc Hoguin
2185198dee
Fix HTTP/2 push
2017-11-30 15:01:01 +01:00
Loïc Hoguin
a66eb5cd6e
Don't send chunked bodies when the method is HEAD
2017-11-29 22:09:33 +01:00
Loïc Hoguin
b0519af42a
Don't include a content-length in 204 responses
2017-11-29 21:38:10 +01:00
Loïc Hoguin
8d9404e9b6
Reject absolute URIs with no authority components
2017-11-29 18:33:50 +01:00
Loïc Hoguin
d50bab8e78
Reject absolute URIs with userinfo components in HTTP/1.1
2017-11-29 18:10:11 +01:00
Loïc Hoguin
aea172857f
Reject whitespace before/after HTTP/1.1 header names properly
2017-11-29 18:01:16 +01:00
Loïc Hoguin
0246d9a4e6
Don't send a GOAWAY frame on close when the h2 preface is invalid
2017-11-29 17:49:42 +01:00
Loïc Hoguin
d33a0df3fe
Ignore stray HTTP/2 stream messages that we expect
2017-11-29 16:41:40 +01:00
Loïc Hoguin
73126e7693
Add many rfc7540 tests, improve detection of malformed requests
2017-11-29 14:54:47 +01:00
Loïc Hoguin
4cdd1aa70e
Add more flow control tests to rfc7540 and fix related issues
2017-11-27 22:49:50 +01:00
Loïc Hoguin
7f80ff28a5
Add more rfc7540 tests along with their respective fixes
2017-11-27 19:00:35 +01:00
Loïc Hoguin
bc82679330
Fix a few rfc7540 tests
...
Cowboy takes a few shortcuts to avoid wasting resources when
there is a protocol error. The RFC wants us to send a different
error depending on the state of the stream at the time of the
error, and for us to maintain the connection in cases where we
would have to spend valuable resources to decode headers. In
all these cases Cowboy will simply close the connection with
an appropriate error.
2017-11-27 13:42:04 +01:00
