gr1ffon/quic-go
1
0
Fork 0
forked from quic-go/quic-go
Code Pull Requests Activity
2,201 Commits 1 Branch 126 Tags
69998c19cbbdcfe98167899c722afdf8c7d0a933
Commit Graph

155 Commits

Author SHA1 Message Date
Marten Seemann
69998c19cb add the offset field to the BLOCKED and STREAM_BLOCKED frames 
The value is not yet set, but parsing and writing works.
 2017-12-14 19:00:06 +07:00
Marten Seemann
03977c1a25 pass RST_STREAM frames directly to the stream 2017-12-14 17:00:41 +07:00
Marten Seemann
2d31440510 pass MAX_STREAM_DATA frames directly to the stream 2017-12-14 16:50:47 +07:00
Marten Seemann
823098d6e2 rename stream.AddStreamFrame to stream.HandleStreamFrame 2017-12-14 16:50:47 +07:00
Marten Seemann
a6e44f3bfc implement parsing and writing of the STOP_SENDING frame 2017-12-13 17:58:50 +07:00
Marten Seemann
73ab97aa95 simplify the stream by directly popping STREAM frames from it 2017-12-12 18:48:08 +07:00
Marten Seemann
7fd0157e1f Merge pull request #1006 from lucas-clemente/more-window-updates 
decrease the threshold to send MAX_{STREAM}_DATA frames
 2017-12-12 18:30:55 +07:00
Marten Seemann
eb3e253be2 send MAX_{STREAM}_DATA frames more frequently 
WINDOW_UPDATEs are relatively small, and it doesn't cost much to grant
the peer more flow control credit earlier.
 2017-12-12 18:06:49 +07:00
Marten Seemann
4aca4d64b7 don't add more STREAM frames to a packet if remaining size is too small 2017-12-12 18:00:05 +07:00
Marten Seemann
4b4e487486 remove the error return value from wire.Frame.MinLength 
No functional change expected.
The error was only non-nil if some required values for the STOP_WAITING
frame were not set. It should be sufficient to throw an error when
attempting to write an invalid STOP_WAITING frame.
 2017-12-12 17:33:04 +07:00
Marten Seemann
632040bd2b Merge pull request #1001 from lucas-clemente/min-packet-size 
implement the minimum packet size requirements
 2017-12-12 16:37:53 +07:00
Marten Seemann
489ea7fa1a Merge pull request #997 from lucas-clemente/stream-refactoring 
refactor sending of stream data
 2017-12-08 22:55:42 +07:00
Marten Seemann
7e627d38ea Merge pull request #999 from lucas-clemente/reject-version-0 
reject packets with version 0 sent by the client
 2017-12-08 22:54:08 +07:00
Marten Seemann
8e8892b064 remove the SentFin method from the stream 
When a FIN is dequeued from the stream by the streamFramer, it is
guaranteed to be sent out. There's no need to explicitely signal that to
the stream.
 2017-12-08 14:05:16 +07:00
Marten Seemann
71af5758e2 remove the ShouldSendFin method from the stream 
GetDataForWriting now has two return parameters: the data and if a FIN
should be sent.
 2017-12-08 14:05:14 +07:00
Marten Seemann
085624be20 replace stream.LenOfDataForWriting by HasDataForWriting 
The return value (the length of data for writing) was only used to
determine if the stream has data for writing. Therefore it's easier to
just return a bool.

No functional change expected.
 2017-12-08 14:04:00 +07:00
Marten Seemann
23ce5a8554 drop packets for new gQUIC connections that are too small 2017-12-08 12:34:59 +07:00
Marten Seemann
6019634286 pad Initial packets to the required minimum size (1200 bytes) 2017-12-08 11:50:43 +07:00
Marten Seemann
f8cbb0ae03 remove unused utils.GetByteOrder 
This function was needed when we supported both Q039 (using big endian
encoding) and ealier versions (using little endian encoding).
 2017-12-08 11:33:40 +07:00
Marten Seemann
d7ceebd644 implement parsing and writing of the new ACK frame 2017-12-08 09:35:55 +07:00
Marten Seemann
11f746a183 implement parsing and writing of the new STREAM frames 2017-12-08 09:20:47 +07:00
Marten Seemann
1a515d1371 implement the new variable integer length frame types 2017-12-08 09:18:54 +07:00
Marten Seemann
759b0d87b1 refactor packet unpacking 
This replaces version.UsesMaxDataFrame by versoin.UsesIETFFrameFormat.
That way, we can have two separate code paths in the unpacker to unpack
either gQUIC frames or IETF frames.
 2017-12-08 09:18:54 +07:00
Marten Seemann
0ac728f96e implement the new variable integer length encoding 2017-12-08 09:18:54 +07:00
Marten Seemann
62b0d99dcd reject packets with version 0 sent by the client 
Version 0 is an invalid version. For packets sent by the server, it is
used to identify the Version Negotiation Packet.
 2017-12-08 09:14:41 +07:00
Marten Seemann
25a6dc9654 implement stateless handling of Initial packets for the TLS server 2017-12-08 08:45:54 +07:00
Marten Seemann
a79e9ee35e parse and write STOP_WAITING frames for packet number 0 2017-12-07 21:33:34 +07:00
Marten Seemann
6ce3336be5 parse and write ACKs that for packet number 0 2017-12-07 21:33:34 +07:00
Marten Seemann
196efa873d Merge pull request #990 from lucas-clemente/fix-989 
don't send a window update after the final offset was received
 2017-12-07 18:55:19 +07:00
Marten Seemann
f51cfe9fe3 don't send a window update after the final offset was received 
Receiving a final offset means the peer is done sending on that stream,
and there's no need to grant additional flow control credit.
 2017-12-07 18:22:39 +07:00
Marten Seemann
7ba613c3b9 use the mint default cookie protector to encrypt and decrypt cookies 2017-12-07 18:20:22 +07:00
Marten Seemann
80a827f133 fix the crypto setup so that it works with the recent mint changes 
mint now supports stateless retries, and requires that the cookie
protector is set in the mint.Config and returns a new alert when a
retry is performed.
 2017-12-07 13:49:45 +07:00
Marten Seemann
8c2404edf5 send a reserved version number in the SHLO 2017-12-06 10:20:22 +07:00
Marten Seemann
1a3852aec6 send a reserved version number in the EncryptedExtensions message 2017-12-06 10:20:22 +07:00
Marten Seemann
be29963637 send a reserved version number in version negotiation packets 2017-12-06 10:20:22 +07:00
Marten Seemann
1f5cd31569 implement a function to get version slices containing reserved versions 2017-12-06 10:20:16 +07:00
Marten Seemann
45e43ada40 implement the recent changes to the version negotiation packet 2017-12-06 08:40:43 +07:00
Marten Seemann
7487dc22bf Merge pull request #977 from lucas-clemente/fix-971 
update validation of version negotiation
 2017-12-06 07:02:49 +07:00
Marten Seemann
b070fcbd43 Merge pull request #976 from lucas-clemente/fix-973 
make initial_max_stream_id optional and parse initial_max_stream_id_uni
 2017-12-06 06:58:15 +07:00
Marten Seemann
ce70ba497d Merge pull request #968 from lucas-clemente/remove-ack-every-20-packets 
remove old code to send an ACK every 20 packets
 2017-12-06 06:57:23 +07:00
Marten Seemann
7c8715616e update validation of version negotiation 
The negotiated_version parameter was recently moved from the
client_hello TLS handshake message to the encrypted_extensions.
 2017-12-05 10:35:04 +07:00
Marten Seemann
c30064bb51 parse TLS extensions containing the initial_max_stream_id_uni parameter 
We're not sending the initial_max_stream_id_uni parameter, which
implicitely sets this value to 0, i.e. the peer is not allowed to open
unidirectional streams.
 2017-12-05 09:17:43 +07:00
Marten Seemann
a72efca38d don't require the initial_max_stream_id in the transport parameters 
The draft was recently changed to make this value optional.
 2017-12-05 08:56:49 +07:00
Marten Seemann
184f81d8c4 remove old code to send an ACK every 20 packets 
This was needed before QUIC 39, and should have been removed when we
dropped support for QUIC 38.
 2017-12-01 12:14:54 +07:00
Marten Seemann
f440a65ef1 move the version field in front of the packet number in the long header 
This was recently changed in the IETF draft.
 2017-11-30 11:34:34 +07:00
Marten Seemann
4076ab587e add a string representation for the Long Header packet types 2017-11-22 15:57:22 -08:00
Jan Rüth
281fec1b2d Make error chan not block for a single write, allowing ParseHandshakeMessage goroutine to gracefully exit 2017-11-22 09:54:03 +01:00
Marten Seemann
683f244054 rename the Cleartext Long Header type to Handshake 
This was recently changed in the draft.
 2017-11-13 10:55:33 +08:00
Marten Seemann
7c3d6abb4b fix parsing of the Header type byte 
In order to determine if a packet is a Version Negotiation Packet, it is
not sufficient to just look at bit 0x1. Other packet types also have
that bit set, e.g. the Retry packet (packet type 0x3). Instead, we have
to look at the last 3 bits. This fix will work as long as IETF QUIC
doesn't define more than 8 long header types.
 2017-11-13 10:23:17 +08:00
Marten Seemann
234d5aab36 echo the offered version in IETF Version Negotiation Packets 2017-11-12 14:37:43 +08:00