3de2727b94
Perf: Prefer more efficient cipher suites
...
Prefer AES128 in order to maximize throughput
2021-03-02 07:01:16 +01:00
82a065ac3e
Make perf up and download more efficient
...
Pass a bigger buffer to write calls to minimize the amount of write calls.
2021-03-02 07:01:16 +01:00
7700046af6
Aggregate perf stats and print them
2021-03-02 07:01:16 +01:00
3bbb15181f
Capture per request timings in perf client
2021-03-02 07:01:16 +01:00
f811d3635c
Allow to specify the perf client to run for a certain amount of time
2021-03-02 07:01:16 +01:00
e8af8fd507
Make handling of uni and bidi streams equal
...
Uni response streams are now accepted as part of the
request lifecycle, which makes it easier to gather stats.
2021-03-02 07:01:16 +01:00
12039602ae
perf: use tokio::try_join instead of select
...
We want all subtasks to complete, and not only a single one.
2021-03-02 07:01:16 +01:00
270016a1da
Use owned buffer API for sending data in the bulk test
...
This avoids the allocation inside the the connection Mutex.
2021-03-02 06:49:34 +01:00
31109bd16f
Make owned bytes write APIs available in quinn
...
This extends the public quinn API to offer support for writing owned buffers.
Besides the universal `write_chunks` API which supports a variable amount
of buffers the API also offers convenience methods for writing either a
single buffer completely or an arbitrary amount of buffers completely.
2021-03-02 06:49:34 +01:00
5ce48e1897
Account for offset=0 requiring no space
...
When a frame contained data for an offset of 0, the transmit logic did not
fully fill a packet. The reason for this is that the logic reserved 1 byte for
writing an offset. However storing offset 0 doesn't require 1 byte,
because it will be encoded with a special flag in the message header.
Without GSO, this isn't a huge issue. It mostly means we are wasting
1 byte per datagram.
However with GSO, this actually leads to data corruption:
When using GSO, and packet isn't fully filled, it later gets padded
to MTU size. Padding a packet ending with a STREAM frame that
doesn't contain a length information is however invalid, and will let
the receiver assume the padding is part of part of the data. This means
the receiver will receive an extra `0` at the end of a stream - and
depending on the duplicate detection at the receiver either the 0
or the correct byte in the next packet will win.
2021-03-01 22:14:43 +01:00
62924202fa
quinn-proto: reorder code from bytes_source module
2021-03-01 13:06:10 -08:00
4abf5a64e0
quinn-proto: merge bytes_source module into connection::streams::send
2021-03-01 13:06:10 -08:00
971265d9ff
quinn-proto: provide datagrams API access through special-purpose type
2021-03-01 13:06:10 -08:00
61129c811e
quinn-proto: move datagram write logic into DatagramState
2021-03-01 13:06:10 -08:00
2637dfe575
quinn-proto: move incoming datagram frame handling into DatagramState
2021-03-01 13:06:10 -08:00
d25ce15239
quinn-proto: move datagram receive logic into DatagramState
2021-03-01 13:06:10 -08:00
c6843a7c84
quinn-proto: derive Default for DatagramState
2021-03-01 13:06:10 -08:00
71484b57aa
quinn-proto: move datagram types into separate module
2021-03-01 13:06:10 -08:00
afe8a6cfee
quinn-proto: inline single-use record_sent_max_data() method
2021-03-01 13:06:10 -08:00
a11abc648e
quinn-proto: inline single-use flow_blocked() method
2021-03-01 13:06:10 -08:00
cc1218ed12
quinn-proto: inline single-use poll_unblocked() method
2021-03-01 13:06:10 -08:00
29e8a914d8
quinn-proto: standardize on ch suffix
2021-03-01 13:06:10 -08:00
5a7b888934
quinn-proto: extract separate RecvStream interface type
2021-03-01 13:06:10 -08:00
0b350e5d11
quinn-proto: extract separate SendStream interface type
2021-03-01 13:06:10 -08:00
8131bcc7b6
quinn-proto: split streams module into two parts
2021-03-01 13:06:10 -08:00
d4bfc25d6a
quinn-proto: move API logic into Streams
2021-03-01 13:06:10 -08:00
8bbe908dbd
quinn-proto: add public Streams interface
2021-03-01 13:06:10 -08:00
584b889494
quinn-proto: rename Streams to StreamsState
2021-03-01 13:06:10 -08:00
dde27bf919
quinn-proto: refactor handling of peer parameters
2021-03-01 13:06:10 -08:00
1b9293366f
quinn-proto: inline single-use congestion_blocked() method
2021-03-01 13:06:10 -08:00
5a7a80ec1e
quinn-proto: move probe queueing logic into PacketSpace
2021-03-01 13:06:10 -08:00
d22800ac79
quinn-proto: move more methods into PacketBuilder
2021-03-01 13:06:10 -08:00
7e0f3fa7ab
quinn-proto: move finish_packet() into PacketBuilder
2021-03-01 13:06:10 -08:00
e1df56f40c
quinn-proto: move PacketBuilder into a separate module
2021-03-01 13:06:10 -08:00
6bddfdea1a
quinn-proto: handle handshake packets separately
2021-03-01 13:06:10 -08:00
816e570151
quinn-proto: inline single-use reject_0rtt() method
2021-03-01 13:06:10 -08:00
18ed973568
quinn-proto: refactor how ACKs are passed to the congestion controller
2021-03-01 13:06:10 -08:00
4e6b8c6fe4
quinn-proto: add comment to clarify need for custom iteration
2021-03-01 13:06:10 -08:00
c4af9ecb1c
fuzz: change config syntax to allow merging
2021-03-01 13:06:10 -08:00
0020e1ee39
Add a Bytes array based send interface to quinn-proto
...
This change modifies quinn-proto to allow callers to submit a list
of one or more owned `Bytes` chunks for transmission instead of
pure byte slices. This will provide a more efficient zero-copy
interface for applications which already make use of owned
`Bytes` buffers.
The internals of quinn-proto have been refactored in order to keep
the ability to pass `&[u8]` buffers and defer the conversion into
`Bytes` as long as possible, in order to avoid unnecessary allocations
if no data can be stored due to flow control.
2021-03-01 09:20:20 +01:00
eb1911d416
Write Bytes into SendBuffer
...
This changes the interface from SendBuffer to directly accept owned
`Bytes` buffers, instead of converting to them.
2021-03-01 09:20:20 +01:00
a9aaf40943
Implement perf protocol for standardized benchmarking
2021-02-28 22:31:43 +01:00
14db88562d
quinn: split streams module in send/recv parts
2021-02-25 12:06:38 -08:00
24cf82ef83
Expose iterator-like read API
2021-02-25 12:06:38 -08:00
dd23094007
quinn-proto: remove unnecessary RecvState::Closed
2021-02-25 12:06:38 -08:00
241cc9cad1
Fix premature abandoment of STOP_SENDING transmission
2021-02-25 16:02:00 +01:00
dd67e91425
Fix state leak for receive streams stopped after being finished
2021-02-25 16:02:00 +01:00
589534ff4e
Tweak method name to avoid confusion with having received FIN
2021-02-25 16:02:00 +01:00
b54e363cff
Fixed typo allow_wpin -> allow_spin
2021-02-25 05:13:00 +01:00
be27d52134
Do not send any packets after a close frame
...
Since the loop was not exited after a close frame was transmitted,
other data in the same space could have been transmitted after
the frame. This adds a `break` to fix this.
2021-02-24 21:04:30 +01:00
Matthias Einwag