QUIC                                                     J. Iyengar, Ed.
Internet-Draft                                                    Google
Intended status: Standards Track                         M. Thomson, Ed.
Expires: September 14, November 22, 2017                                       Mozilla
                                                          March 13,
                                                            May 21, 2017

           QUIC: A UDP-Based Multiplexed and Secure Transport
                      draft-ietf-quic-transport-02
                      draft-ietf-quic-transport-03

Abstract

   This document defines the core of the QUIC transport protocol.  This
   document describes connection establishment, packet format,
   multiplexing and reliability.  Accompanying documents describe the
   cryptographic handshake and loss detection.

Note to Readers

   Discussion of this draft takes place on the QUIC working group
   mailing list (quic@ietf.org), which is archived at
   https://mailarchive.ietf.org/arch/search/?email_list=quic .

   Working Group information can be found at https://github.com/quicwg ;
   source code and issues list for this draft can be found at
   https://github.com/quicwg/base-drafts/labels/transport .

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at http://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   This Internet-Draft will expire on September 14, November 22, 2017.

Copyright Notice

   Copyright (c) 2017 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   4
   2.  Conventions and Definitions . . . . . . . . . . . . . . . . .   4   5
     2.1.  Notational Conventions  . . . . . . . . . . . . . . . . .   5
   3.  A QUIC Overview . . . . . . . . . . . . . . . . . . . . . . .   5   6
     3.1.  Low-Latency Connection Establishment  . . . . . . . . . .   6
     3.2.  Stream Multiplexing . . . . . . . . . . . . . . . . . . .   6
     3.3.  Rich Signaling for Congestion Control and Loss Recovery .   6   7
     3.4.  Stream and Connection Flow Control  . . . . . . . . . . .   6   7
     3.5.  Authenticated and Encrypted Header and Payload  . . . . .   7
     3.6.  Connection Migration and Resilience to NAT Rebinding  . .   7   8
     3.7.  Version Negotiation . . . . . . . . . . . . . . . . . . .   8
   4.  Versions  . . . . . . . . . . . . . . . . . . . . . . . . . .   8
   5.  Packet Types and Formats  . . . . . . . . . . . . . . . . . .   8   9
     5.1.  Long Header . . . . . . . . . . . . . . . . . . . . . . .   9
     5.2.  Short Header  . . . . . . . . . . . . . . . . . . . . . .  11
     5.3.  Version Negotiation Packet  . . . . . . . . . . . . . . .  12  13
     5.4.  Cleartext Packets . . . . . . . . . . . . . . . . . . . .  13  14
       5.4.1.  Client Initial Packet . . . . . . . . . . . . . . . .  14
       5.4.2.  Server Stateless Retry Packet . . . . . . . . . . . .  15
       5.4.3.  Server Cleartext Packet . . . . . . . . . . . . . . .  15
       5.4.4.  Client Cleartext Packet . . . . . . . . . . . . . . .  16
     5.5.  Encrypted  Protected Packets . . . . . . . . . . . . . . . . . . . .  14  16
     5.6.  Public Reset Packet . . . . . . . . . . . . . . . . . . .  15  17
       5.6.1.  Public Reset Proof  . . . . . . . . . . . . . . . . .  15  18
     5.7.  Connection ID . . . . . . . . . . . . . . . . . . . . . .  16  18
     5.8.  Packet Numbers  . . . . . . . . . . . . . . . . . . . . .  16  18
       5.8.1.  Initial Packet Number . . . . . . . . . . . . . . . .  17  19
     5.9.  Handling Packets from Different Versions  . . . . . . . .  17  20
   6.  Frames and Frame Types  . . . . . . . . . . . . . . . . . . .  18  20
   7.  Life of a Connection  . . . . . . . . . . . . . . . . . . . .  19  22
     7.1.  Version Negotiation . . . . . . . . . . . . . . . . . . .  19  23
       7.1.1.  Using Reserved Versions . . . . . . . . . . . . . . .  20  24
     7.2.  Cryptographic and Transport Handshake . . . . . . . . . .  21  24
     7.3.  Transport Parameters  . . . . . . . . . . . . . . . . . .  22  25
       7.3.1.  Transport Parameter Definitions . . . . . . . . . . .  24  27
       7.3.2.  Values of Transport Parameters for 0-RTT  . . . . . .  24  27
       7.3.3.  New Transport Parameters  . . . . . . . . . . . . . .  25  28
       7.3.4.  Version Negotiation Validation  . . . . . . . . . . .  25  28
     7.4.  Stateless Retries . . . . . . . . . . . . . . . . . . . .  29
     7.5.  Proof of Source Address Ownership . . . . . . . . . . . .  27
       7.4.1.  30
       7.5.1.  Client Address Validation Procedure . . . . . . . . .  27
       7.4.2.  31
       7.5.2.  Address Validation on Session Resumption  . . . . . .  28
       7.4.3.  31
       7.5.3.  Address Validation Token Integrity  . . . . . . . . .  29
     7.5.  32
     7.6.  Connection Migration  . . . . . . . . . . . . . . . . . .  29
     7.6.  32
       7.6.1.  Privacy Implications of Connection Migration  . . . .  33
       7.6.2.  Address Validation for Migrated Connections . . . . .  34
     7.7.  Connection Termination  . . . . . . . . . . . . . . . . .  30  34
   8.  Frame Types and Formats . . . . . . . . . . . . . . . . . . .  31  35
     8.1.  STREAM Frame  . . . . . . . . . . . . . . . . . . . . . .  31  35
     8.2.  ACK Frame . . . . . . . . . . . . . . . . . . . . . . . .  32  37
       8.2.1.  ACK Block Section . . . . . . . . . . . . . . . . . .  34  39
       8.2.2.  Timestamp Section . . . . . . . . . . . . . . . . . .  35  40
       8.2.3.  ACK Frames and Packet Protection  . . . . . . . . . .  37  41
     8.3.  WINDOW_UPDATE  MAX_DATA Frame  . . . . . . . . . . . . . . . . . . .  38
     8.4.  BLOCKED Frame . . . .  42
     8.4.  MAX_STREAM_DATA Frame . . . . . . . . . . . . . . . . . .  39  43
     8.5.  RST_STREAM  MAX_STREAM_ID Frame . . . . . . . . . . . . . . . . . . . .  39  44
     8.6.  PADDING  BLOCKED Frame . . . . . . . . . . . . . . . . . . . . . .  40  44
     8.7.  PING frame  STREAM_BLOCKED Frame  . . . . . . . . . . . . . . . . . .  44
     8.8.  STREAM_ID_NEEDED Frame  . . . . .  40
     8.8.  CONNECTION_CLOSE frame . . . . . . . . . . . .  45
     8.9.  RST_STREAM Frame  . . . . .  40
     8.9.  GOAWAY Frame . . . . . . . . . . . . . . .  45
     8.10. PADDING Frame . . . . . . .  41
   9.  Packetization and Reliability . . . . . . . . . . . . . . .  46
     8.11. PING frame  .  42
     9.1.  Special Considerations for PMTU Discovery . . . . . . . .  44
   10. Streams: QUIC's Data Structuring Abstraction . . . . . . . .  45
     10.1.  Life of a Stream . . . . . .  46
     8.12. NEW_CONNECTION_ID Frame . . . . . . . . . . . . . .  45
       10.1.1.  idle . . .  46
     8.13. CONNECTION_CLOSE frame  . . . . . . . . . . . . . . . . .  47
     8.14. GOAWAY Frame  . . . .  47
       10.1.2.  open . . . . . . . . . . . . . . . . . .  48
   9.  Packetization and Reliability . . . . . .  47
       10.1.3.  half-closed (local) . . . . . . . . . .  49
     9.1.  Special Considerations for PMTU Discovery . . . . . .  48
       10.1.4.  half-closed (remote) . .  51
   10. Streams: QUIC's Data Structuring Abstraction  . . . . . . . .  51
     10.1.  Stream Identifiers . . . . . .  48
       10.1.5.  closed . . . . . . . . . . . . .  52
     10.2.  Life of a Stream . . . . . . . . . .  48
     10.2.  Stream Identifiers . . . . . . . . . .  52
       10.2.1.  idle . . . . . . . . .  50
     10.3.  Stream Concurrency . . . . . . . . . . . . . . .  54
       10.2.2.  open . . . .  50
     10.4.  Sending and Receiving Data . . . . . . . . . . . . . . .  51
     10.5.  Stream Prioritization . . . . .  54
       10.2.3.  half-closed (local)  . . . . . . . . . . . .  51
   11. Flow Control . . . .  55
       10.2.4.  half-closed (remote) . . . . . . . . . . . . . . . .  55
       10.2.5.  closed . . . .  52
     11.1.  Edge Cases and Other Considerations . . . . . . . . . .  54
       11.1.1.  Mid-stream RST_STREAM . . . . . . . . .  56
     10.3.  Stream Concurrency . . . . . .  54
       11.1.2.  Response to a RST_STREAM . . . . . . . . . . . . .  56
     10.4.  Sending and Receiving Data .  54
       11.1.3.  Offset Increment . . . . . . . . . . . . . .  57
     10.5.  Stream Prioritization  . . . .  54
       11.1.4.  BLOCKED frames . . . . . . . . . . . . .  57
   11. Flow Control  . . . . . .  55
   12. Error Handling . . . . . . . . . . . . . . . . . .  58
     11.1.  Edge Cases and Other Considerations  . . . . .  55
     12.1.  Connection Errors . . . . .  59
       11.1.1.  Response to a RST_STREAM . . . . . . . . . . . . . .  55
     12.2.  Stream Errors  60
       11.1.2.  Data Limit Increments  . . . . . . . . . . . . . . .  60
     11.2.  Stream Limit Increment . . . . . .  56
     12.3.  Error Codes . . . . . . . . . . .  61
       11.2.1.  Blocking on Flow Control . . . . . . . . . . .  56
   13. Security and Privacy Considerations . . .  61
     11.3.  Stream Final Offset  . . . . . . . . . .  60
     13.1.  Spoofed ACK Attack . . . . . . . .  61
   12. Error Handling  . . . . . . . . . . .  60
   14. IANA Considerations . . . . . . . . . . . .  62
     12.1.  Connection Errors  . . . . . . . . .  61
     14.1.  QUIC Transport Parameter Registry . . . . . . . . . .  62
     12.2.  Stream Errors  .  61
   15. References . . . . . . . . . . . . . . . . . . . .  63
     12.3.  Error Codes  . . . . .  62
     15.1.  Normative References . . . . . . . . . . . . . . . . .  63
   13. Security and Privacy Considerations .  62
     15.2.  Informative References . . . . . . . . . . . .  67
     13.1.  Spoofed ACK Attack . . . . .  63
     15.3.  URIs . . . . . . . . . . . . . .  67
     13.2.  Slowloris Attacks  . . . . . . . . . . . .  64
   Appendix A.  Contributors . . . . . . .  68
     13.3.  Stream Fragmentation and Reassembly Attacks  . . . . . .  68
     13.4.  Stream Commitment Attack . . . . . . .  64
   Appendix B.  Acknowledgments . . . . . . . . .  68
   14. IANA Considerations . . . . . . . . .  64
   Appendix C.  Change Log . . . . . . . . . . . .  69
     14.1.  QUIC Transport Parameter Registry  . . . . . . . . .  64
     C.1.  Since draft-ietf-quic-transport-01: . .  69
   15. References  . . . . . . . . .  64 . . . . . . . . . . . . . . . .  70
     15.1.  Normative References . . . . . . . . . . . . . . . . . .  70
     15.2.  Informative References . . . . . . . . . . . . . . . . .  71
     15.3.  URIs . . . . . . . . . . . . . . . . . . . . . . . . . .  72
   Appendix A.  Contributors . . . . . . . . . . . . . . . . . . . .  72
   Appendix B.  Acknowledgments  . . . . . . . . . . . . . . . . . .  72
   Appendix C.  Change Log . . . . . . . . . . . . . . . . . . . . .  72
     C.1.  Since draft-ietf-quic-transport-02  . . . . . . . . . . .  72
     C.2.  Since draft-ietf-quic-transport-00: draft-ietf-quic-transport-01  . . . . . . . . . . .  66  73
     C.3.  Since draft-hamilton-quic-transport-protocol-01: draft-ietf-quic-transport-00  . . . .  67 . . . . . . .  75
     C.4.  Since draft-hamilton-quic-transport-protocol-01 . . . . .  76
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  67  76

1.  Introduction

   QUIC is a multiplexed and secure transport protocol that runs on top
   of UDP.  QUIC aims to provide a flexible set of features that allow
   it to be a general-purpose transport for multiple applications.

   QUIC implements techniques learned from experience with TCP, SCTP and
   other transport protocols.  Using UDP as the substrate, QUIC seeks to
   be compatible with legacy clients and middleboxes.  QUIC
   authenticates all of its headers and encrypts most of the data it
   exchanges, including its signaling.  This allows the protocol to
   evolve without incurring a dependency on upgrades to middleboxes.
   This document describes the core QUIC protocol, including the
   conceptual design, wire format, and mechanisms of the QUIC protocol
   for connection establishment, stream multiplexing, stream and
   connection-level flow control, and data reliability.

   Accompanying documents describe QUIC's loss detection and congestion
   control [QUIC-RECOVERY], and the use of TLS 1.3 for key negotiation
   [QUIC-TLS].

2.  Conventions and Definitions

   The words "MUST", "MUST NOT", "SHOULD", and "MAY" are used in this
   document.  It's not shouting; when they are capitalized, they have
   the special meaning defined in [RFC2119].

   Definitions of terms that are used in this document:

   Client:  The endpoint initiating a QUIC connection.

   Server:  The endpoint accepting incoming QUIC connections.

   Endpoint:  The client or server end of a connection.

   Stream:  A logical, bi-directional channel of ordered bytes within a
      QUIC connection.

   Connection:  A conversation between two QUIC endpoints with a single
      encryption context that multiplexes streams within it.

   Connection ID:  The identifier for a QUIC connection.

   QUIC packet:  A well-formed UDP payload that can be parsed by a QUIC
      receiver.  QUIC packet size in this document refers to the UDP
      payload size.

2.1.  Notational Conventions

   Packet and frame diagrams use the format described in [RFC2360]
   Section 3.1, with the following additional conventions:

   [x]  Indicates that x is optional

   {x}  Indicates that x is encrypted

   x (A)  Indicates that x is A bits long

   x (A/B/C) ...  Indicates that x is one of A, B, or C bits long

   x (*) ...  Indicates that x is variable-length

3.  A QUIC Overview

   This section briefly describes QUIC's key mechanisms and benefits.
   Key strengths of QUIC include:

   o  Low-latency connection establishment

   o  Multiplexing without head-of-line blocking

   o  Authenticated and encrypted header and payload

   o  Rich signaling for congestion control and loss recovery

   o  Stream and connection flow control

   o  Connection migration and resilience to NAT rebinding

   o  Version negotiation

3.1.  Low-Latency Connection Establishment

   QUIC relies on a combined cryptographic and transport handshake for
   setting up a secure transport connection.  QUIC connections are
   expected to commonly use 0-RTT handshakes, meaning that for most QUIC
   connections, data can be sent immediately following the client
   handshake packet, without waiting for a reply from the server.  QUIC
   provides a dedicated stream (Stream ID 1) 0) to be used for performing
   the cryptographic handshake and QUIC options negotiation.  The format
   of the QUIC options and parameters used during negotiation are
   described in this document, but the handshake protocol that runs on
   Stream ID 1 0 is described in the accompanying cryptographic handshake
   draft [QUIC-TLS].

3.2.  Stream Multiplexing

   When application messages are transported over TCP, independent
   application messages can suffer from head-of-line blocking.  When an
   application multiplexes many streams atop TCP's single-bytestream
   abstraction, a loss of a TCP segment results in blocking of all
   subsequent segments until a retransmission arrives, irrespective of
   the application streams that are encapsulated in subsequent segments.
   QUIC ensures that lost packets carrying data for an individual stream
   only impact that specific stream.  Data received on other streams can
   continue to be reassembled and delivered to the application.

3.3.  Rich Signaling for Congestion Control and Loss Recovery

   QUIC's packet framing and acknowledgments carry rich information that
   help both congestion control and loss recovery in fundamental ways.
   Each QUIC packet carries a new packet number, including those
   carrying retransmitted data.  This obviates the need for a separate
   mechanism to distinguish acknowledgments for retransmissions from
   those for original transmissions, avoiding TCP's retransmission
   ambiguity problem.  QUIC acknowledgments also explicitly encode the
   delay between the receipt of a packet and its acknowledgment being
   sent, and together with the monotonically-increasing packet numbers,
   this allows for precise network roundtrip-time (RTT) calculation.
   QUIC's ACK frames support up to 256 ACK blocks, so QUIC is more
   resilient to reordering than TCP with SACK support, as well as able
   to keep more bytes on the wire when there is reordering or loss.

3.4.  Stream and Connection Flow Control

   QUIC implements stream- and connection-level flow control, closely
   following HTTP/2's flow control mechanisms. control.  At a high
   level, a QUIC receiver advertises the absolute byte offset within each stream up to
   which the receiver maximum amount of data that it
   is willing to receive data. on each stream.  As data is sent, received, and
   delivered on a particular stream, the receiver sends
   WINDOW_UPDATE MAX_STREAM_DATA
   frames that increase the advertised offset limit for that stream, allowing
   the peer to send more data on that stream.

   In addition to this stream-level flow control, QUIC implements
   connection-level flow control to limit the aggregate buffer that a
   QUIC receiver is willing to allocate to all streams on a connection.
   Connection-level flow control works in the same way as stream-level
   flow control, but the bytes delivered and highest received offset the limits are
   all aggregates aggregated
   across all streams.

3.5.  Authenticated and Encrypted Header and Payload

   TCP headers appear in plaintext on the wire and are not
   authenticated, causing a plethora of injection and header
   manipulation issues for TCP, such as receive-window manipulation and
   sequence-number overwriting.  While some of these are mechanisms used
   by middleboxes to improve TCP performance, others are active attacks.
   Even "performance-enhancing" middleboxes that routinely interpose on
   the transport state machine end up limiting the evolvability of the
   transport protocol, as has been observed in the design of MPTCP
   [RFC6824] and in its subsequent deployability issues.

   Generally, QUIC packets are always authenticated and the payload is
   typically fully encrypted.  The parts of the packet header which are
   not encrypted are still authenticated by the receiver, so as to
   thwart any packet injection or manipulation by third parties.  Some
   early handshake packets, such as the Version Negotiation packet, are
   not encrypted, but information sent in these unencrypted handshake
   packets is later verified as part of cryptographic processing.

   PUBLIC_RESET packets that reset a connection are currently not
   authenticated.

3.6.  Connection Migration and Resilience to NAT Rebinding

   QUIC connections are identified by a 64-bit Connection ID, randomly
   generated by the client.  QUIC's consistent connection ID allows
   connections to survive changes to the client's IP and port, such as
   those caused by NAT rebindings or by the client changing network
   connectivity to a new address.  QUIC provides automatic cryptographic
   verification of a rebound client, since the client continues to use
   the same session key for encrypting and decrypting packets.  The
   consistent connection ID can be used to allow migration of the
   connection to a new server IP address as well, since the Connection
   ID remains consistent across changes in the client's and the server's
   network addresses.

3.7.  Version Negotiation

   QUIC version negotiation allows for multiple versions of the protocol
   to be deployed and used concurrently.  Version negotiation is
   described in Section 7.1.

4.  Versions

   QUIC versions are identified using a 32-bit value.

   The version 0x00000000 is reserved to represent an invalid version.
   This version of the specification is identified by the number
   0x00000001.

   Version 0x000000001 of QUIC uses TLS as a cryptographic handshake
   protocol, as described in [QUIC-TLS].

   Versions with the most significant 16 bits of the version number
   cleared are reserved for use in future IETF consensus documents.

   Versions that follow the pattern 0x?a?a?a?a are reserved for use in
   forcing version negotiation to be exercised.  That is, any version
   number where the low four bits of all octets is 1010 (in binary).  A
   client or server MAY advertise support for any of these reserved
   versions.

   Reserved version numbers will probably never represent a real
   protocol; a client MAY use one of these version numbers with the
   expectation that the server will initiate version negotiation; a
   server MAY advertise support for one of these versions and can expect
   that clients ignore the value.

   [[RFC editor: please remove the remainder of this section before
   publication.]]

   The version number for the final version of this specification
   (0x00000001), is reserved for the version of the protocol that is
   published as an RFC.

   Version numbers used to identify IETF drafts are created by adding
   the draft number to 0xff000000.  For example, draft-ietf-quic-
   transport-13 would be identified as 0xff00000D.

   Implementors are encouraged to register version numbers of QUIC that
   they are using for private experimentation on the github wiki [4].

5.  Packet Types and Formats

   We first describe QUIC's packet types and their formats, since some
   are referenced in subsequent mechanisms.

   All numeric values are encoded in network byte order (that is, big-
   endian) and all field sizes are in bits.  When discussing individual
   bits of fields, the least significant bit is referred to as bit 0.
   Hexadecimal notation is used for describing the value of fields.

   Any QUIC packet has either a long or a short header, as indicated by
   the Header Form bit.  Long headers are expected to be used early in
   the connection before version negotiation and establishment of 1-RTT
   keys, and for public resets.  Short headers are minimal version-
   specific headers, which can be used after version negotiation and
   1-RTT keys are established.

5.1.  Long Header
    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+
   |1|   Type (7)  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                       Connection ID (64)                      +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                       Packet Number (32)                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                         Version (32)                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Payload (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                       Figure 1: Long Header Format

   Long headers are used for packets that are sent prior to the
   completion of version negotiation and establishment of 1-RTT keys.
   Once both conditions are met, a sender SHOULD switch to sending
   short-form headers.  While inefficient, long headers MAY be used for
   packets encrypted with 1-RTT keys.  The long form allows for special
   packets, such as the Version Negotiation and the Public Reset packets
   to be represented in this uniform fixed-length packet format.  A long
   header contains the following fields:

   Header Form:  The most significant bit (0x80) of the first octet is
      set to 1 for long headers and 0 for short headers.

   Long Packet Type:  The remaining seven bits of first octet of a long
      packet is the packet type.  This field can indicate one of 128
      packet types.  The types specified for this version are listed in
      Table 1.

   Connection ID:  Octets 1 through 8 contain the connection ID.
      Section 5.7 describes the use of this field in more detail.

   Packet Number:  Octets 9 to 12 contain the packet number.  {{packet-
      numbers} describes the use of packet numbers.

   Version:  Octets 13 to 16 contain the selected protocol version.
      This field indicates which version of QUIC is in use and
      determines how the rest of the protocol fields are interpreted.

   Payload:  Octets from 17 onwards (the rest of QUIC packet) are the
      payload of the packet.

   The following packet types are defined:

          +------+-------------------------------+-------------+

         +------+-------------------------------+---------------+
         | Type | Name                          | Section       |
          +------+-------------------------------+-------------+
         +------+-------------------------------+---------------+
         | 01   | Version Negotiation           | Section 5.3   |
         |      |                               |               |
         | 02   | Client Cleartext Initial                | Section 5.4 5.4.1 |
         |      |                               |               |
         | 03   | Non-Final Server Cleartext Stateless Retry        | Section 5.4 5.4.2 |
         |      |                               |               |
         | 04   | Final Server Cleartext              | Section 5.4 5.4.3 |
         |      |                               |               |
         | 05   | 0-RTT Encrypted Client Cleartext              | Section 5.5 5.4.4 |
         |      |                               |               |
         | 06   | 1-RTT Encrypted (key phase 0) 0-RTT Protected               | Section 5.5   |
         |      |                               |               |
         | 07   | 1-RTT Encrypted Protected (key phase 1) 0) | Section 5.5   |
         |      |                               |               |
         | 08   | 1-RTT Protected (key phase 1) | Section 5.5   |
         |      |                               |               |
         | 09   | Public Reset                  | Section 5.6   |
          +------+-------------------------------+-------------+
         +------+-------------------------------+---------------+

                     Table 1: Long Header Packet Types

   The header form, packet type, connection ID, packet number and
   version fields of a long header packet are version-independent.  The
   types of packets defined in Table 1 are version-specific.  See
   Section 5.9 for details on how packets from different versions of
   QUIC are interpreted.

   (TODO: Should the list of packet types be version-independent?)

   The interpretation of the fields and the payload are specific to a
   version and packet type.  Type-specific semantics for this version
   are described in Section 5.3, Section 5.6, Section 5.4, and
   Section 5.5. the following sections.

5.2.  Short Header
    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+
   |0|C|K| Type (5)|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                     [Connection ID (64)]                      +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      Packet Number (8/16/32)                ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Encrypted                     Protected Payload (*)                   ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                       Figure 2: Short Header Format

   The short header can be used after the version and 1-RTT keys are
   negotiated.  This header form has the following fields:

   Header Form:  The most significant bit (0x80) of the first octet of a
      packet is the header form.  This bit is set to 0 for the short
      header.

   Connection ID Flag:  The second bit (0x40) of the first octet
      indicates whether the Connection ID field is present.  If set to
      1, then the Connection ID field is present; if set to 0, the
      Connection ID field is omitted.

   Key Phase Bit:  The third bit (0x20) of the first octet indicates the
      key phase, which allows a recipient of a packet to identify the
      packet protection keys that are used to protect the packet.  See
      [QUIC-TLS] for details.

   Short Packet Type:  The remaining 5 bits of the first octet include
      one of 32 packet types.  Table 2 lists the types that are defined
      for short packets.

   Connection ID:  If the Connection ID Flag is set, a connection ID
      occupies octets 1 through 8 of the packet.  See Section 5.7 for
      more details.

   Packet Number:  The length of the packet number field depends on the
      packet type.  This field can be 1, 2 or 4 octets long depending on
      the short packet type.

   Encrypted

   Protected Payload:  Packets with a short header always include a
      1-RTT protected payload.

   The packet type in a short header currently determines only the size
   of the packet number field.  Additional types can be used to signal
   the presence of other fields.

                       +------+--------------------+
                       | Type | Packet Number Size |
                       +------+--------------------+
                       | 01   | 1 octet            |
                       |      |                    |
                       | 02   | 2 octets           |
                       |      |                    |
                       | 03   | 4 octets           |
                       +------+--------------------+

                    Table 2: Short Header Packet Types

   The header form, connection ID flag and connection ID of a short
   header packet are version-independent.  The remaining fields are
   specific to the selected QUIC version.  See Section 5.9 for details
   on how packets from different versions of QUIC are interpreted.

5.3.  Version Negotiation Packet

   A Version Negotiation packet has long headers with a type value of
   0x01 and is sent only by servers and servers.  The Version Negotiation packet is
   a response to a client packet of an unsupported version.  It uses that contains a
   long header and contains:

   o  Octet 0: 0x81

   o  Octets 1-8: Connection version that is not
   supported by the server.

   The connection ID (echoed)

   o  Octets 9-12: Packet Number (echoed)

   o  Octets 13-16: field contains a server-selected connection ID that
   the client MUST use for subsequent packets, see Section 5.7.

   The packet number and version fields echo corresponding values from
   the triggering client packet.  This allows clients some assurance
   that the server received the packet and that the Version (echoed)

   o  Octets 17+: Payload Negotiation
   packet was not carried in a packet with a spoofed source address.

   The payload of the Version Negotiation packet is a list of 32-bit
   versions which the server supports, as shown below.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Supported Version 1 (32)                 ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                   [Supported Version 2 (32)]                ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                  ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                   [Supported Version N (32)]                ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                   Figure 3: Version Negotiation Packet

   See Section 7.1 for a description of the version negotiation process.

5.4.  Cleartext Packets

   Cleartext packets are sent during the handshake prior to key
   negotiation.  A Client Cleartext packet contains:

   o  Octet 0: 0x82

   o  Octets 1-8: Connection ID (initial)

   o  Octets 9-12: Packet number

   o  Octets 13-16: Version

   o  Octets 17+: Payload

   Non-Final Server Cleartext

   All cleartext packets contain:

   o  Octet 0: 0x83

   o  Octets 1-8: Connection ID (echoed)

   o  Octets 9-12: Packet Number

   o  Octets 13-16: Version

   o  Octets 17+: Payload

   Final Server Cleartext contain the current QUIC version in the version
   field.

   The payload of cleartext packets contains:

   o  Octet 0: 0x84

   o  Octets 1-8: Connection ID (final)
   o  Octets 9-12: also includes an integrity check,
   which is described in [QUIC-TLS].

5.4.1.  Client Initial Packet Number

   o  Octets 13-16: Version

   o  Octets 17+: Payload

   The client MUST choose Client Initial packet uses long headers with a random 64-bit type value and use it as the
   initial Connection ID in all packets until of
   0x02.  It carries the server replies with first cryptographic handshake message sent by
   the final Connection ID. client.

   The server echoes client populates the client's Connection connection ID in Non-Final Server Cleartext packets.  The first Final Server
   Cleartext and all subsequent packets MUST use the final Connection
   ID, as described in Section 5.7.

   The payload of field with randomly selected
   values, unless it has received a Cleartext packet consists of from the server.  If the
   client has received a sequence of frames,
   as described in Section 6.

   (TODO: Add hash before frames.)

5.5.  Encrypted Packets

   Packets encrypted with either 0-RTT or 1-RTT keys may be sent with
   long headers.  Different packet types explicitly indicate from the
   encryption level server, the connection ID field
   uses the value provided by the server.

   The packet number used for ease of decryption.  These Client Initial packets contain:

   o  Octet 0: 0x85, 0x86 or 0x87

   o  Octets 1-8: Connection ID (initial or final)

   o  Octets 9-12: Packet Number

   o  Octets 13-16: Version

   o  Octets 17+: Encrypted Payload

   A first octet of 0x85 indicates is initialized with
   a 0-RTT packet.  After random value each time the 1-RTT keys
   are established, key phases new contents are used by created for the packet.
   Retransmissions of the QUIC packet protection to
   identify contents increment the correct packet protection keys.  The initial key phase
   is 0.  See [QUIC-TLS] for more details. number by
   one, see (Section 5.8).

   The encrypted payload is both authenticated and encrypted using
   packet protection keys.  [QUIC-TLS] describes packet protection in
   detail.  After decryption, the plaintext consists of a sequence of
   frames, as described in Section 6.

5.6.  Public Reset Packet

   A Public Reset Client Initial packet is only sent by servers and is used to abruptly
   terminate communications.  Public Reset is provided as an option consists of
   last resort a STREAM frame (or
   frames) for stream 0 containing a server that does not have access cryptographic handshake message,
   plus any PADDING frames necessary to ensure that the state of a
   connection.  This packet is intended at
   least the minimum size (see Section 9).  This stream frame always
   starts at an offset of 0 (see Section 7.4).

   The client uses the Client Initial Packet type for use by a server any packet that has lost state
   (for example, through
   contains an initial cryptographic handshake message.  This includes
   all cases where a crash or outage).  A server that wishes new packet containing the initial cryptographic
   message needs to
   communicate a fatal connection error MUST use be created, this includes the packets sent after
   receiving a CONNECTION_CLOSE
   frame if it has sufficient state to do so. Version Negotiation (Section 5.3) or Server Stateless
   Retry packet (Section 5.4.2).

5.4.2.  Server Stateless Retry Packet

   A Public Reset Server Stateless Retry packet contains:

   o  Octet 0: 0x88

   o  Octets 1-8: Echoed data (octets 1-8 of received packet)

   o  Octets 9-12: Echoed data (octets 9-12 uses long headers with a type value
   of received packet)

   o  Octets 13-16: Version

   o  Octets 17+: Public Reset Proof

   For 0x03.  It carries cryptographic handshake messages and
   acknowledgments.  It is used by a client server that sends wishes to perform a
   stateless retry (see Section 7.4).

   The connection ID on every packet, the
   Connection ID field is simply an echo of the initial Connection ID,
   and the Packet Number field includes an echo of the client's in a Server Stateless Retry packet contains a
   server selected value, see Section 5.7.

   The packet number (and, depending on field echoes the client's packet number length, 0, 2, or
   3 additional octets from of the client's packet).

   A Public Reset packet sent by triggering
   client packet.  This allows a server indicates client to verify that it does not
   have the state necessary to continue with server
   received its packet.

   After receiving a connection.  In this
   case, Server Stateless Retry packet, the server will include client uses a
   new Client Initial packet containing the fields that prove that it
   originally participated in next cryptographic handshake
   message.  The client retains the connection (see Section 5.6.1 for
   details).

   Upon receipt state of its cryptographic
   handshake, but discards all transport state.  In effect, the next
   cryptographic handshake message is sent on a Public Reset new connection.  The new
   Client Initial packet that contains is sent in a valid proof, packet with a newly randomized
   packet number and starting at a stream offset of 0.

   Continuing the cryptographic handshake is necessary to ensure that an
   attacker cannot force a downgrade of any cryptographic parameters.
   In addition to continuing the cryptographic handshake, the client
   MUST tear down state associated with remember the connection. results of any version negotiation that occurred
   (see Section 7.1).  The client MAY also retain any observed RTT or
   congestion state that it has accumulated for the flow, but other
   transport state MUST then cease sending packets on be discarded.

   The payload of the connection Server Stateless Retry packet contains STREAM
   frames and SHOULD
   discard any subsequent packets that arrive.  A Public Reset that does
   not could contain PADDING and ACK frames.  A server can only
   send a valid proof MUST be ignored.

5.6.1.  Public Reset Proof

   TODO: Details single Server Stateless Retry packet in response to be added.

5.7.  Connection ID

   QUIC connections are identified by their 64-bit Connection ID.  All each
   Client Initial packet that is receives.

5.4.3.  Server Cleartext Packet

   A Server Cleartext packet uses long headers contain with a Connection ID.  Short headers indicate the
   presence type value of a Connection ID using the CONNECTION_ID flag.  When
   present, the Connection ID
   0x04.  It is in used to carry acknowledgments and cryptographic
   handshake messages from the same location server.

   The connection ID field in all a Server Cleartext packet
   headers, making it straightforward for middleboxes, such as load
   balancers, to locate and use it.

   When contains a
   connection is initiated, the client MUST choose a random value
   and use it as the initial Connection ID until the final value is
   available.  The initial Connection ID that is a suggestion to chosen by the server.
   The server echoes this value in all packets until the handshake is
   successful (see [QUIC-TLS]).  On a successful handshake, the server
   MUST select the final Connection ID for the connection and use it in
   Final Section 5.7).

   The first Server Cleartext packets.  This final Connection ID MAY be the
   one proposed by the client or MAY be packet contains a new server-selected value.
   All randomized packet
   number.  This value is increased for each subsequent packets from packet sent by
   the server MUST contain as described in Section 5.8.

   The payload of this value.  On
   handshake completion, the client MUST switch to using the final
   Connection ID for all subsequent packets.

   Thus, all Client Cleartext packets, 0-RTT Encrypted packets, packet contains STREAM frames and Non-
   Final Server Cleartext packets MUST use the client's randomly-
   generated initial Connection ID.  Final Server Cleartext packets,
   1-RTT Encrypted packets, could contain
   PADDING and all short-header packets MUST use the
   final Connection ID.

5.8. ACK frames.

5.4.4.  Client Cleartext Packet Numbers

   The

   A Client Cleartext packet number is uses long headers with a 64-bit unsigned number type value of
   0x05, and is used as part of sent when the client has received a cryptographic nonce for Server Cleartext
   packet encryption.  Each endpoint maintains from the server.

   The connection ID field in a separate Client Cleartext packet number for sending and receiving. contains a
   server-selected connection ID, see Section 5.7.

   The Client Cleartext packet includes a packet number for sending MUST increase by at least that is one after sending any
   packet.

   A QUIC endpoint MUST NOT reuse a packet number within the same
   connection (that is, under the same cryptographic keys).  If
   higher than the last Client Initial, 0-RTT Protected or Client
   Cleartext packet that was sent.  The packet number for sending reaches 2^64 - 1, the sender MUST close the
   connection by sending a CONNECTION_CLOSE frame with the error code
   QUIC_SEQUENCE_NUMBER_LIMIT_REACHED (connection termination is
   described in incremented for
   each subsequent packet, see Section 7.6.)

   To reduce the number 5.8.

   The payload of bits required to represent the this packet number
   over the wire, only contains STREAM frames and could contain
   PADDING and ACK frames.

5.5.  Protected Packets

   Packets that are protected with 0-RTT keys are sent with long
   headers.  Packets that are protected with 1-RTT keys MAY be sent with
   long headers.  The different packet types explicitly indicate the least significant bits of
   encryption level and therefore the packet number keys that are transmitted over the wire, up used to 32 bits.  The actual remove
   packet
   number protection.

   Packets protected with 0-RTT keys use a type value of 0x06.  The
   connection ID field for each a 0-RTT packet is reconstructed at the receiver based on selected by the
   largest packet number client.

   The client can send 0-RTT packets after having received on a successfully authenticated
   packet.

   A packet number is decoded by finding from
   the packet number value server if that is
   closest to the next expected packet.  The next expected packet is does not complete the
   highest received packet number plus one.  For example, handshake.  Even if
   the highest
   successfully authenticated packet had a packet number of 0xaa82f30e,
   then a packet containing client receives a 16-bit value of 0x1f94 will be decoded as
   0xaa831f94.

   The sender MUST use a packet number size able to represent more than
   twice as large a range than the difference between the largest
   acknowledged packet and packet number being sent.  A peer receiving
   the packet will then correctly decode different connection ID from the packet number, unless server, it
   MUST NOT update the
   packet is delayed in transit such that connection ID it arrives after many higher-
   numbered packets have been received.  An endpoint MAY uses for 0-RTT packets.  This
   enables consistent routing for all 0-RTT packets.

   Packets protected with 1-RTT keys that use long headers use a larger
   packet number size to safeguard against such reordering.

   As a result, the size type
   value of the packet number encoding is at least one 0x07 for key phase 0 and 0x08 for key phase 1; see
   [QUIC-TLS] for more than the base 2 logarithm of details on the number use of contiguous
   unacknowledged key phases.  The connection
   ID field for these packet numbers, including types MUST contain the new packet.

   For example, if an endpoint has received an acknowledgment value selected by
   the server, see Section 5.7.

   The version field for protected packets is the current QUIC version.

   The packet
   0x6afa2f, sending a packet with a number of 0x6b4264 requires field contains a
   16-bit or larger packet number encoding; whereas a 32-bit number, which increases
   with each packet
   number sent, see Section 5.8 for details.

   The payload is needed to send a protected using authenticated encryption.  [QUIC-TLS]
   describes packet with protection in detail.  After decryption, the
   plaintext consists of a number sequence of 0x6bc107.

5.8.1.  Initial frames, as described in
   Section 6.

5.6.  Public Reset Packet Number

   The initial value for packet number MUST be a 31-bit random number.
   That is, the value

   A Public Reset packet is selected from an uniform random distribution
   between 0 and 2^31-1.  [RFC4086] provides guidance on the generation
   of random values.

   The first set of packets only sent by servers and is used to abruptly
   terminate communications.  Public Reset is provided as an endpoint MUST include the low
   32-bits option of
   last resort for a server that does not have access to the packet number.  Once any packet state of a
   connection.  This is intended for use by a server that has been acknowledged,
   subsequent packets can lost state
   (for example, through a crash or outage).  A server that wishes to
   communicate a fatal connection error MUST use a shorter CONNECTION_CLOSE
   frame if it has sufficient state to do so.

   A Public Reset packet uses long headers with a type value of 0x09.

   The connection ID and packet number encoding.

5.9.  Handling Packets from Different Versions

   Between different versions the following things are guaranteed to
   remain constant:

   o  the location of fields together contain octets
   1 through 12 from the header form flag,

   o  the location of packet that triggered the Connection reset.  For a client
   that sends a connection ID flag in short headers,

   o  the location and size of on every packet, the Connection ID field in both header
      forms,

   o  the location and size
   is simply an echo of the Version field in long headers, and
   o  the location client's Connection ID, and size of the Packet
   Number field in long headers.

   Implementations MUST assume that an unsupported version uses includes an
   unknown packet format.  All other fields MUST be ignored when
   processing a echo of the client's packet that contains an unsupported version.

6.  Frames and Frame Types

   The number.
   Depending on the client's packet number length it might also include
   0, 2, or 3 additional octets from the protected payload of cleartext packets and the plaintext after decryption
   of encrypted payloads consists of client
   packet.

   The version field contains the current QUIC version.

   A Public Reset packet sent by a sequence of frames, as shown server indicates that it does not
   have the state necessary to continue with a connection.  In this
   case, the server will include the fields that prove that it
   originally participated in
   Figure 4.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame 1 (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame 2 (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                  ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame N (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                  Figure 4: Contents the connection (see Section 5.6.1 for
   details).

   Upon receipt of Encrypted Payload

   Encrypted payloads a Public Reset packet that contains a valid proof, a
   client MUST contain at least one frame, tear down state associated with the connection.  The
   client MUST then cease sending packets on the connection and MAY SHOULD
   discard any subsequent packets that arrive.  A Public Reset that does
   not contain
   multiple frames and multiple frame types.

   Frames MUST fit within a single QUIC packet and valid proof MUST NOT span a be ignored.

5.6.1.  Public Reset Proof

   TODO: Details to be added.

5.7.  Connection ID

   QUIC
   packet boundary.  Each frame begins with a Frame Type byte,
   indicating its type, followed by additional type-dependent fields:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Type (8)    |           Type-Dependent Fields (*)         ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                      Figure 5: Generic Frame Layout

   Frame types connections are listed identified by their 64-bit Connection ID.  All
   long headers contain a Connection ID.  Short headers indicate the
   presence of a Connection ID using the CONNECTION_ID flag.  When
   present, the Connection ID is in Table 3.  Note that the Frame Type byte same location in
   STREAM and ACK frames is used to carry other frame-specific flags.
   For all other frames, packet
   headers, making it straightforward for middleboxes, such as load
   balancers, to locate and use it.

   The client MUST choose a random connection ID and use it in Client
   Initial packets (Section 5.4.1).  If the Frame Type byte simply identifies client has received any
   packet from the
   frame.  These frames are explained server, it uses the connection ID it received from
   the server.

   When the server receives a Client Initial packet, it chooses a new
   value for the connection ID and sends that in more detail as they are
   referenced later its response.  The
   server MUST send a new connection ID in any packet that is sent in
   response to a Client Initial packet.  This includes Version
   Negotiation (Section 5.3), Server Stateless Retry (Section 5.4.2),
   and the first Server Cleartext packet (Section 5.4.3).  The server
   MAY choose to use the document.

           +------------------+------------------+-------------+
           | Type-field value | Frame type       | Definition  |
           +------------------+------------------+-------------+
           | 0x00             | PADDING          | Section 8.6 |
           |                  |                  |             |
           | 0x01             | RST_STREAM       | Section 8.5 |
           |                  |                  |             |
           | 0x02             | CONNECTION_CLOSE | Section 8.8 |
           |                  |                  |             |
           | 0x03             | GOAWAY           | Section 8.9 |
           |                  |                  |             |
           | 0x04             | WINDOW_UPDATE    | Section 8.3 |
           |                  |                  |             |
           | 0x05             | BLOCKED          | Section 8.4 |
           |                  |                  |             |
           | 0x07             | PING             | Section 8.7 |
           |                  |                  |             |
           | 0x40 - 0x7f      | ACK              | Section 8.2 |
           |                  |                  |             |
           | 0x80 - 0xff      | STREAM           | Section 8.1 |
           +------------------+------------------+-------------+

                           Table 3: Frame Types

7.  Life of a Connection that the client initially selects.

   A QUIC connection is server MUST NOT propose a single conversation between two QUIC
   endpoints.  QUIC's different connection establishment intertwines version
   negotiation with ID in response to a
   Client Cleartext packet (Section 5.4.4).  A Client Cleartext packet
   is only sent after the cryptographic and transport handshakes server has committed to reduce maintaining connection establishment latency,
   state.

5.8.  Packet Numbers

   The packet number is a 64-bit unsigned number and is used as described in Section 7.2.  Once
   established, part of
   a connection may migrate to cryptographic nonce for packet encryption.  Each endpoint maintains
   a different IP or port separate packet number for sending and receiving.  The packet
   number for sending MUST increase by at
   either endpoint, due to NAT rebinding or mobility, as described in least one after sending any
   packet, unless otherwise specified (see Section 7.5.  Finally 5.8.1).

   A QUIC endpoint MUST NOT reuse a packet number within the same
   connection may be terminated by either
   endpoint, as described in Section 7.6.

7.1.  Version Negotiation

   QUIC's connection establishment begins with version negotiation,
   since all communication between (that is, under the same cryptographic keys).  If the endpoints, including
   packet and
   frame formats, relies on number for sending reaches 2^64 - 1, the sender MUST close the two endpoints agreeing on a version.

   A QUIC
   connection begins with a client by sending a handshake packet.
   The details of CONNECTION_CLOSE frame with the handshake mechanisms are error code
   QUIC_SEQUENCE_NUMBER_LIMIT_REACHED (connection termination is
   described in Section 7.2,
   but all of the initial packets sent from 7.7.)

   To reduce the client number of bits required to represent the server
   MUST use packet number
   over the long header format and MUST specify wire, only the version least significant bits of the
   protocol being used.

   When the server receives a packet from a client with the long header
   format, it compares number
   are transmitted over the client's version wire, up to 32 bits.  The actual packet
   number for each packet is reconstructed at the versions it supports.

   If receiver based on the version selected
   largest packet number received on a successfully authenticated
   packet.

   A packet number is decoded by finding the client packet number value that is not acceptable
   closest to the
   server, next expected packet.  The next expected packet is the server discards
   highest received packet number plus one.  For example, if the incoming highest
   successfully authenticated packet and responds with had a
   Version Negotiation packet (Section 5.3).  This includes number of 0xaa82f30e,
   then a list packet containing a 16-bit value of
   versions that the server 0x1f94 will accept.  A server be decoded as
   0xaa831f94.

   The sender MUST send use a Version
   Negotiation packet for every packet that it receives with an
   unacceptable version.

   If the packet contains a version that is acceptable number size able to represent more than
   twice as large a range than the server, difference between the server proceeds with largest
   acknowledged packet and packet number being sent.  A peer receiving
   the handshake (Section 7.2).  This commits packet will then correctly decode the server to packet number, unless the version
   packet is delayed in transit such that the client selected.

   When the client receives it arrives after many higher-
   numbered packets have been received.  An endpoint MAY use a Version Negotiation larger
   packet from the
   server, it should select an acceptable protocol version.  If the
   server lists an acceptable version, the client selects that version
   and reattempts number size to create safeguard against such reordering.

   As a connection using that version.  Though result, the
   contents size of a packet might not change in response to version
   negotiation, a client MUST increase the packet number it uses on
   every encoding is at least one
   more than the base 2 logarithm of the number of contiguous
   unacknowledged packet it sends.  Packets MUST continue to use long headers and
   MUST include numbers, including the new negotiated protocol version.

   The client MUST use the long header format and include its selected
   version on all packets until it has 1-RTT keys and it packet.

   For example, if an endpoint has received an acknowledgment for packet
   0x6afa2f, sending a packet from the server which is not with a Version Negotiation packet.

   A client MUST NOT change the version it uses unless it number of 0x6b4264 requires a
   16-bit or larger packet number encoding; whereas a 32-bit packet
   number is in response needed to send a packet with a number of 0x6bc107.

   Version Negotiation packet from (Section 5.3), Server Stateless Retry
   (Section 5.4.2), and Public Reset (Section 5.6) packets have special
   rules for populating the server.  Once a client
   receives a packet number field.

5.8.1.  Initial Packet Number

   The initial value for packet number MUST be selected from an uniform
   random distribution between 0 and 2^31-1.  That is, the server which is not a Version Negotiation
   packet, it MUST ignore Version Negotiation packets lower 31 bits
   of the packet number are randomized.  [RFC4086] provides guidance on
   the same
   connection.

   Version negotiation uses unprotected data. generation of random values.

   The result first set of the
   negotiation packets sent by an endpoint MUST be revalidated as part include the low
   32-bits of the cryptographic
   handshake (see Section 7.3.4).

7.1.1.  Using Reserved Versions

   For a server to packet number.  Once any packet has been acknowledged,
   subsequent packets can use a new version in the future, clients must
   correctly handle unsupported versions.  To help ensure this, a server
   SHOULD include a reserved version (see Section 4) while generating shorter packet number encoding.

   A client that receives a Version Negotiation packet.

   The design of version negotiation permits (Section 5.3) or Server
   Stateless Retry packet (Section 5.4.2) MUST generate a server to avoid
   maintaining state new initial
   packet number.  This ensures that the first transmission attempt for
   a Client Initial packet (Section 5.4.1) always contains a randomized
   packet number, but packets that it rejects in this fashion.
   However, when contain retransmissions increment the server generates a Version Negotiation packet, it
   cannot randomly
   packet number.

   A client MUST NOT generate a reserved version number.  This is because new initial packet number if it discards
   the server is required to include packet.  This might happen if the same value in information the client
   retransmits its transport
   parameters (see Section 7.3.4).  To avoid Client Initial packet.

5.9.  Handling Packets from Different Versions

   Between different versions the selected version number
   changing during connection establishment, following things are guaranteed to
   remain constant:

   o  the reserved version SHOULD
   be generated as a function location of values that will be available to the
   server when later generating its handshake packets.

   A pseudorandom function that takes client address information (IP header form flag,

   o  the location of the Connection ID flag in short headers,

   o  the location and
   port) size of the Connection ID field in both header
      forms,

   o  the location and size of the client selected version as input would ensure that
   there is sufficient variability Version field in long headers, and

   o  the values location and size of the Packet Number field in long headers.

   Implementations MUST assume that a server uses.

   A client MAY send a packet using a reserved an unsupported version number.  This can uses an
   unknown packet format.  All other fields MUST be used to solicit a list of supported versions from ignored when
   processing a server.

7.2.  Cryptographic packet that contains an unsupported version.

6.  Frames and Transport Handshake

   QUIC relies on a combined cryptographic Frame Types

   The payload of cleartext packets and transport handshake to
   minimize connection establishment latency.  QUIC allocates stream 1
   for the cryptographic handshake.  This version plaintext after decryption
   of QUIC uses TLS 1.3
   [QUIC-TLS].

   QUIC provides this stream with reliable, ordered delivery protected payloads consists of data.
   In return, the cryptographic handshake provides QUIC with:

   o  authenticated key exchange, where

      *  a server is always authenticated,

      * a client is optionally authenticated,

      *  every connection produces distinct and unrelated keys,

      *  keying material is usable for packet protection for both 0-RTT
         and 1-RTT packets, and

      *  1-RTT keys have forward secrecy

   o  authenticated values for the transport parameters of the peer (see
      Section 7.3)

   o  authenticated confirmation sequence of version negotiation (see
      Section 7.3.4)

   o  authenticated negotiation frames, as shown in
   Figure 4.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame 1 (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame 2 (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                  ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Frame N (*)                        ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                  Figure 4: Contents of an application protocol (TLS uses
      ALPN [RFC7301] for this purpose)

   o  for the server, the ability to carry data that provides assurance
      that the client can receive packets that are addressed with the
      transport address that is claimed by the client (see Section 7.4)

   The initial cryptographic handshake message Protected Payload

   Protected payloads MUST be sent in a single
   packet.  Any second attempt that is triggered by address validation contain at least one frame, and MAY contain
   multiple frames and multiple frame types.

   Frames MUST also be sent fit within a single packet.  This avoids having to
   reassemble a message from multiple packets.  Reassembling messages
   requires that a server maintain state prior to establishing a
   connection, exposing the server to a denial of service risk.

   The first client QUIC packet of the cryptographic handshake protocol and MUST
   fit within NOT span a 1280 octet QUIC packet.  This includes overheads that
   reduce the
   packet boundary.  Each frame begins with a Frame Type byte,
   indicating its type, followed by additional type-dependent fields:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Type (8)    |           Type-Dependent Fields (*)         ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                      Figure 5: Generic Frame Layout

   Frame types are listed in Table 3.  Note that the Frame Type byte in
   STREAM and ACK frames is used to carry other frame-specific flags.
   For all other frames, the Frame Type byte simply identifies the
   frame.  These frames are explained in more detail as they are
   referenced later in the document.

            +-------------+-------------------+--------------+
            | Type Value  | Frame Type Name   | Definition   |
            +-------------+-------------------+--------------+
            | 0x00        | PADDING           | Section 8.10 |
            |             |                   |              |
            | 0x01        | RST_STREAM        | Section 8.9  |
            |             |                   |              |
            | 0x02        | CONNECTION_CLOSE  | Section 8.13 |
            |             |                   |              |
            | 0x03        | GOAWAY            | Section 8.14 |
            |             |                   |              |
            | 0x04        | MAX_DATA          | Section 8.3  |
            |             |                   |              |
            | 0x05        | MAX_STREAM_DATA   | Section 8.4  |
            |             |                   |              |
            | 0x06        | MAX_STREAM_ID     | Section 8.5  |
            |             |                   |              |
            | 0x07        | PING              | Section 8.11 |
            |             |                   |              |
            | 0x08        | BLOCKED           | Section 8.6  |
            |             |                   |              |
            | 0x09        | STREAM_BLOCKED    | Section 8.7  |
            |             |                   |              |
            | 0x0a        | STREAM_ID_NEEDED  | Section 8.8  |
            |             |                   |              |
            | 0x0b        | NEW_CONNECTION_ID | Section 8.12 |
            |             |                   |              |
            | 0xa0 - 0xbf | ACK               | Section 8.2  |
            |             |                   |              |
            | 0xc0 - 0xff | STREAM            | Section 8.1  |
            +-------------+-------------------+--------------+

                           Table 3: Frame Types

7.  Life of a Connection

   A QUIC connection is a single conversation between two QUIC
   endpoints.  QUIC's connection establishment intertwines version
   negotiation with the cryptographic and transport handshakes to reduce
   connection establishment latency, as described in Section 7.2.  Once
   established, a connection may migrate to a different IP or port at
   either endpoint, due to NAT rebinding or mobility, as described in
   Section 7.6.  Finally a connection may be terminated by either
   endpoint, as described in Section 7.7.

7.1.  Version Negotiation

   QUIC's connection establishment begins with version negotiation,
   since all communication between the endpoints, including packet and
   frame formats, relies on the two endpoints agreeing on a version.

   A QUIC connection begins with a client sending a handshake packet.
   The details of the handshake mechanisms are described in Section 7.2,
   but all of the initial packets sent from the client to the server
   MUST use the long header format and MUST specify the version of the
   protocol being used.

   When the server receives a packet from a client with the long header
   format, it compares the client's version to the versions it supports.

   If the version selected by the client is not acceptable to the
   server, the server discards the incoming packet and responds with a
   Version Negotiation packet (Section 5.3).  This includes a list of
   versions that the server will accept.

   A server sends a Version Negotiation packet for every packet that it
   receives with an unacceptable version.  This allows a server to
   process packets with unsupported versions without retaining state.
   Though either the initial client packet or the version negotiation
   packet that is sent in response could be lost, the client will send
   new packets until it successfully receives a response.

   If the packet contains a version that is acceptable to the server,
   the server proceeds with the handshake (Section 7.2).  This commits
   the server to the version that the client selected.

   When the client receives a Version Negotiation packet from the
   server, it should select an acceptable protocol version.  If the
   server lists an acceptable version, the client selects that version
   and reattempts to create a connection using that version.  Though the
   contents of a packet might not change in response to version
   negotiation, a client MUST increase the packet number it uses on
   every packet it sends.  Packets MUST continue to use long headers and
   MUST include the new negotiated protocol version.

   The client MUST use the long header format and include its selected
   version on all packets until it has 1-RTT keys and it has received a
   packet from the server which is not a Version Negotiation packet.

   A client MUST NOT change the version it uses unless it is in response
   to a Version Negotiation packet from the server.  Once a client
   receives a packet from the server which is not a Version Negotiation
   packet, it MUST ignore other Version Negotiation packets on the same
   connection.  Similarly, a client MUST ignore a Version Negotiation
   packet if it has already received and acted on a Version Negotiation
   packet.

   A client MUST ignore a Version Negotiation packet that lists the
   client's chosen version.

   Version negotiation uses unprotected data.  The result of the
   negotiation MUST be revalidated as part of the cryptographic
   handshake (see Section 7.3.4).

7.1.1.  Using Reserved Versions

   For a server to use a new version in the future, clients must
   correctly handle unsupported versions.  To help ensure this, a server
   SHOULD include a reserved version (see Section 4) while generating a
   Version Negotiation packet.

   The design of version negotiation permits a server to avoid
   maintaining state for packets that it rejects in this fashion.
   However, when the server generates a Version Negotiation packet, it
   cannot randomly generate a reserved version number.  This is because
   the server is required to include the same value in its transport
   parameters (see Section 7.3.4).  To avoid the selected version number
   changing during connection establishment, the reserved version SHOULD
   be generated as a function of values that will be available to the
   server when later generating its handshake packets.

   A pseudorandom function that takes client address information (IP and
   port) and the client selected version as input would ensure that
   there is sufficient variability in the values that a server uses.

   A client MAY send a packet using a reserved version number.  This can
   be used to solicit a list of supported versions from a server.

7.2.  Cryptographic and Transport Handshake

   QUIC relies on a combined cryptographic and transport handshake to
   minimize connection establishment latency.  QUIC allocates stream 0
   for the cryptographic handshake.  Version 0x00000001 of QUIC uses TLS
   1.3 as described in [QUIC-TLS]; a different QUIC version number could
   indicate that a different cryptographic handshake protocol is in use.

   QUIC provides this stream with reliable, ordered delivery of data.
   In return, the cryptographic handshake provides QUIC with:

   o  authenticated key exchange, where
      *  a server is always authenticated,

      *  a client is optionally authenticated,

      *  every connection produces distinct and unrelated keys,

      *  keying material is usable for packet protection for both 0-RTT
         and 1-RTT packets, and

      *  1-RTT keys have forward secrecy

   o  authenticated values for the transport parameters of the peer (see
      Section 7.3)

   o  authenticated confirmation of version negotiation (see
      Section 7.3.4)

   o  authenticated negotiation of an application protocol (TLS uses
      ALPN [RFC7301] for this purpose)

   o  for the server, the ability to carry data that provides assurance
      that the client can receive packets that are addressed with the
      transport address that is claimed by the client (see Section 7.5)

   The initial cryptographic handshake message MUST be sent in a single
   packet.  Any second attempt that is triggered by address validation
   MUST also be sent within a single packet.  This avoids having to
   reassemble a message from multiple packets.  Reassembling messages
   requires that a server maintain state prior to establishing a
   connection, exposing the server to a denial of service risk.

   The first client packet of the cryptographic handshake protocol MUST
   fit within a 1232 octet QUIC packet payload.  This includes overheads
   that reduce the space available to the cryptographic handshake
   protocol.

   Details of how TLS is integrated with QUIC is provided in more detail
   in [QUIC-TLS].

7.3.  Transport Parameters

   During connection establishment, both endpoints make authenticated
   declarations of their transport parameters.  These declarations are
   made unilaterally by each endpoint.  Endpoints are required to comply
   with the restrictions implied by these parameters; the description of
   each parameter includes rules for its handling.

   The format of the transport parameters is the TransportParameters
   struct from Figure 6.  This is described using the presentation
   language from Section 3 of [I-D.ietf-tls-tls13].

      uint32 QuicVersion;

      enum {
         stream_fc_offset(0),
         connection_fc_offset(1),
         concurrent_streams(2),
         initial_max_stream_data(0),
         initial_max_data(1),
         initial_max_stream_id(2),
         idle_timeout(3),
         truncate_connection_id(4),
         (65535)
      } TransportParameterId;

      struct {
         TransportParameterId parameter;
         opaque value<0..2^16-1>;
      } TransportParameter;

      struct {
         select (Handshake.msg_type) {
            case client_hello:
               QuicVersion negotiated_version;
               QuicVersion initial_version;

            case encrypted_extensions:
               QuicVersion supported_versions<2..2^8-4>;
         };
         TransportParameter parameters<30..2^16-1>;
      } TransportParameters;

                Figure 6: Definition of TransportParameters

   The "extension_data" field of the quic_transport_parameters extension
   defined in [QUIC-TLS] contains a TransportParameters value.  TLS
   encoding rules are therefore used to encode the transport parameters.

   QUIC encodes transport parameters into a sequence of octets, which
   are then included in the cryptographic handshake.  Once the handshake
   completes, the transport parameters declared by the peer are
   available.  Each endpoint validates the value provided by its peer.
   In particular, version negotiation MUST be validated (see
   Section 7.3.4) before the connection establishment is considered
   properly complete.

   Definitions for each of the defined transport parameters are included
   in Section 7.3.1.

7.3.1.  Transport Parameter Definitions

   An endpoint MUST include the following parameters in its encoded
   TransportParameters:

   stream_fc_offset

   initial_max_stream_data (0x0000):  The initial stream level flow control
      offset maximum data
      parameter contains the initial value for the maximum data that can
      be sent on any newly created stream.  This parameter is encoded as
      an unsigned 32-bit integer in units of octets.  The sender of this parameter indicates that the flow
      control offset for all stream data sent toward it  This is this value.

   connection_fc_offset equivalent
      to an implicit MAX_STREAM_DATA frame (Section 8.4) being sent on
      all streams immediately after opening.

   initial_max_data (0x0001):  The connection level flow control
      offset initial maximum data parameter
      contains the initial connection flow control
      window value for the maximum amount of data that can
      be sent on the connection.  This parameter is encoded as an
      unsigned 32-bit integer in units of 1024 octets.  That is, the
      value here is multiplied by 1024 to determine the actual flow control offset.  The sender of this
      parameter sets the byte offset for connection level flow control
      to this maximum
      value.  This is equivalent to sending a WINDOW_UPDATE MAX_DATA (Section 8.3) for
      the connection immediately after completing the handshake.

   concurrent_streams

   initial_max_stream_id (0x0002):  The initial maximum number of concurrent
      streams stream ID
      parameter is contains the initial maximum stream number the peer may
      initiate, encoded as an unsigned 32-bit integer.  This is
      equivalent to sending a MAX_STREAM_ID (Section 8.5) immediately
      after completing the handshake.

   idle_timeout (0x0003):  The idle timeout is a value in seconds that
      is encoded as an unsigned 16-bit integer.  The maximum value is
      600 seconds (10 minutes).

   An endpoint MAY use the following transport parameters:

   truncate_connection_id (0x0004):  The truncated connection identifier
      parameter indicates that packets sent to the peer can omit the
      connection ID.  This can be used by an endpoint where the 5-tuple
      is sufficient to identify a connection.  This parameter is zero
      length.  Omitting the parameter indicates that the endpoint relies
      on the connection ID being present in every packet.

7.3.2.  Values of Transport Parameters for 0-RTT

   Transport parameters from the server SHOULD be remembered by the
   client for use with 0-RTT data.  A client that doesn't remember
   values from a previous connection can instead assume the following
   values: stream_fc_offset (65535), connection_fc_offset (65535),
   concurrent_streams (10), idle_timeout (600), truncate_connection_id
   (absent).

   If assumed values change as a result of completing the handshake, the
   client is expected to respect the new values.  This introduces some
   potential problems, particularly with respect to transport parameters
   that establish limits:

   o  A client might exceed a newly declared connection or stream flow
      control limit MUST be remembered by the client
   for use with 0-RTT data.  If this occurs, the client ceases
      transmission as though the flow control limit was reached.  Once
      WINDOW_UPDATE frames indicating an increase to TLS NewSessionTicket message
   includes the affected flow
      control offsets is received, quic_transport_parameters extension, then those values
   are used for the client can recommence sending.

   o  Similarly, server values when establishing a client might exceed new connection
   using that ticket.  Otherwise, the concurrent stream limit
      declared by transport parameters that the server.
   server advertises during connection establishment are used.

   A client MUST reset any streams server can remember the transport parameters that
      exceed this limit. it advertised, or
   store an integrity-protected copy of the values in the ticket and
   recover the information when accepting 0-RTT data.  A server SHOULD reset any streams it cannot
      handle with a code that allows uses the client to retry any application
      action bound
   transport parameters in determining whether to those streams. accept 0-RTT data.

   A server MAY close a connection if remembered or assumed accept 0-RTT and subsequently provide different values
   for transport parameters cannot be supported, using an error code that for use in the new connection.  If 0-RTT
   data is
   appropriate to accepted by the specific condition.  For example, a
   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA server, the server MUST NOT reduce any limits
   or alter any values that might be used to indicate
   that exceeding flow control limits caused violated by the error.  A client with its
   0-RTT data.  In particular, a server that
   has accepts 0-RTT data MUST NOT
   set values for initial_max_data or initial_max_stream_data that are
   smaller than the remembered value of those parameters.  Similarly, a connection closed due to an error condition SHOULD
   server MUST NOT attempt reduce the value of initial_max_stream_id.

   A server MUST reject 0-RTT when attempting to create data or even abort a new connection. handshake if the
   implied values for transport parameters cannot be supported.

7.3.3.  New Transport Parameters

   New transport parameters can be used to negotiate new protocol
   behavior.  An endpoint MUST ignore transport parameters that it does
   not support.  Absence of a transport parameter therefore disables any
   optional protocol feature that is negotiated using the parameter.

   The definition of a transport parameter SHOULD include a default
   value that a client can use when establishing a new connection.  If
   no default is specified, the value can be assumed to be absent when
   attempting 0-RTT.

   New transport parameters can be registered according to the rules in
   Section 14.1.

7.3.4.  Version Negotiation Validation

   The transport parameters include three fields that encode version
   information.  These retroactively authenticate the version
   negotiation (see Section 7.1) that is performed prior to the
   cryptographic handshake.

   The cryptographic handshake provides integrity protection for the
   negotiated version as part of the transport parameters (see
   Section 7.3).  As a result, modification of version negotiation
   packets by an attacker can be detected.

   The client includes two fields in the transport parameters:

   o  The negotiated_version is the version that was finally selected
      for use.  This MUST be identical to the value that is on the
      packet that carries the ClientHello.  A server that receives a
      negotiated_version that does not match the version of QUIC of QUIC that is
      in use MUST terminate the connection with a
      QUIC_VERSION_NEGOTIATION_MISMATCH error code.

   o  The initial_version is the version that the client initially
      attempted to use.  If the server did not send a version
      negotiation packet Section 5.3, this will be identical to the
      negotiated_version.

   A server that processes all packets in a stateful fashion can
   remember how version negotiation was performed and validate the
   initial_version value.

   A server that does not maintain state for every packet it receives
   (i.e., a stateless server) uses a different process.  If the initial
   and negotiated versions are the same, a stateless server can accept
   the value.

   If the initial version is different from the negotiated_version, a
   stateless server MUST check that it would have sent a version
   negotiation packet if it had received a packet with the indicated
   initial_version.  If a server would have accepted the version
   included in the initial_version and the value differs from the value
   of negotiated_version, the server MUST terminate the connection with
   a QUIC_VERSION_NEGOTIATION_MISMATCH error.

   The server includes a list of versions that it would send in any
   version negotiation packet (Section 5.3) in supported_versions.  This
   value is set even if it did not send a version negotiation packet.

   The client can validate that the negotiated_version is included in
   the supported_versions list and - if version negotiation was
   performed - that is
      in use it would have selected the negotiated version.  A
   client MUST terminate the connection with a
   QUIC_VERSION_NEGOTIATION_MISMATCH error code.

   o  The initial_version is code if the version that
   negotiated_version value is not included in the supported_versions
   list.  A client initially
      attempted to use.  If the server did not send MUST terminate with a
   QUIC_VERSION_NEGOTIATION_MISMATCH error code if version negotiation packet Section 5.3, this will be identical to
   occurred but it would have selected a different version based on the
      negotiated_version.
   value of the supported_versions list.

7.4.  Stateless Retries

   A server that processes all packets in a stateful fashion can
   remember how version negotiation was performed and validate the
   initial_version value. process an initial cryptographic handshake messages from
   a client without committing any state.  This allows a server to
   perform address validation (Section 7.5, or to defer connection
   establishment costs.

   A server that does not maintain generates a response to an initial packet without
   retaining connection state for every MUST use the Server Stateless Retry packet it receives
   (i.e., a stateless server) uses
   (Section 5.4.2).  This packet causes a different process.  If the initial client to reset its transport
   state and negotiated versions are to continue the same, a stateless server can accept connection attempt with new connection
   state while maintaining the value.

   If state of the initial version cryptographic handshake.

   A server MUST NOT send multiple Server Stateless Retry packets in
   response to a client handshake packet.  Thus, any cryptographic
   handshake message that is different from sent MUST fit within a single packet.

   In TLS, the negotiated_version, Server Stateless Retry packet type is used to carry the
   HelloRetryRequest message.

7.5.  Proof of Source Address Ownership

   Transport protocols commonly spend a
   stateless server MUST check round trip checking that a
   client owns the transport address (IP and port) that it would have claims.
   Verifying that a client can receive packets sent to its claimed
   transport address protects against spoofing of this information by
   malicious clients.

   This technique is used primarily to avoid QUIC from being used for
   traffic amplification attack.  In such an attack, a version
   negotiation packet if it had received is sent to
   a packet server with the indicated
   initial_version. spoofed source address information that identifies a
   victim.  If a server would have accepted the version
   included generates more or larger packets in response to
   that packet, the initial_version and the value differs from the value
   of negotiated_version, attacker can use the server MUST terminate to send more data toward
   the connection with
   a QUIC_VERSION_NEGOTIATION_MISMATCH error.

   The server includes a list of versions that victim than it would be able to send on its own.

   Several methods are used in any
   version negotiation QUIC to mitigate this attack.  Firstly,
   the initial handshake packet (Section 5.3) in supported_versions.  This
   value is set even if it did not padded to at least 1280 octets.  This
   allows a server to send a version negotiation packet.

   The client can validate similar amount of data without risking
   causing an amplification attack toward an unproven remote address.

   A server eventually confirms that a client has received its messages
   when the negotiated_version is included in cryptographic handshake successfully completes.  This might
   be insufficient, either because the supported_versions list and - if version negotiation was
   performed - that server wishes to avoid the
   computational cost of completing the handshake, or it would have selected might be that
   the negotiated version.  A
   client MUST terminate size of the connection with a
   QUIC_VERSION_NEGOTIATION_MISMATCH error code if packets that are sent during the
   negotiated_version value handshake is not included in too
   large.  This is especially important for 0-RTT, where the supported_versions
   list.  A client MUST terminate with server
   might wish to provide application data traffic - such as a
   QUIC_VERSION_NEGOTIATION_MISMATCH error code if version negotiation
   occurred but it would have selected response
   to a different version based on request - in response to the
   value of data carried in the supported_versions list.

7.4.  Proof of Source Address Ownership

   Transport protocols commonly spend a round trip checking early data from
   the client.

   To send additional data prior to completing the cryptographic
   handshake, the server then needs to validate that a the client owns the transport
   address (IP and port) that it claims.
   Verifying that a client can receive packets sent to its claimed
   transport

   Source address protects against spoofing validation is therefore performed during the
   establishment of this information by
   malicious clients.

   This technique a connection.  TLS provides the tools that support
   the feature, but basic validation is used primarily to avoid performed by the core transport
   protocol.

7.5.1.  Client Address Validation Procedure

   QUIC from being used for
   traffic amplification attack.  In such an attack, uses token-based address validation.  Any time the server wishes
   to validate a packet client address, it provides the client with a token.
   As long as the token cannot be easily guessed (see Section 7.5.3), if
   the client is sent able to
   a return that token, it proves to the server with spoofed source address information that identifies a
   victim.  If
   it received the token.

   During the processing of the cryptographic handshake messages from a server generates more or larger packets in response
   client, TLS will request that QUIC make a decision about whether to
   proceed based on the information it has.  TLS will provide QUIC with
   any token that packet, was provided by the attacker client.  For an initial packet,
   QUIC can use the server decide to send more data toward abort the victim than connection, allow it would be able to send on its own.

   Several methods are used in proceed, or
   request address validation.

   If QUIC decides to mitigate this attack.  Firstly, request address validation, it provides the initial
   cryptographic handshake packet from with a client token.  The contents of this token are
   consumed by the server that generates the token, so there is padded to at least 1280
   octets.  This allows no need
   for a single well-defined format.  A token could include information
   about the claimed client address (IP and port), a timestamp, and any
   other supplementary information the server will need to send a similar amount of data
   without risking causing an amplication attack toward an unproven
   remote address. validate the
   token in the future.

   The cryptographic handshake is responsible for enacting validation by
   sending the address validation token to the client.  A server eventually confirms that a legitimate
   client has received its messages will include a copy of the token when it attempts to continue
   the handshake.  The cryptographic handshake successfully completes.  This might
   be insufficient, extracts the token then
   asks QUIC a second time whether the token is acceptable.  In
   response, QUIC can either because abort the server wishes connection or permit it to avoid
   proceed.

   A connection MAY be accepted without address validation - or with
   only limited validation - but a server SHOULD limit the
   computational cost data it sends
   toward an unvalidated address.  Successful completion of completing the handshake, or it might be
   cryptographic handshake implicitly provides proof that the size of the client has
   received packets that are sent during from the handshake is too
   large.  This server.

7.5.2.  Address Validation on Session Resumption

   A server MAY provide clients with an address validation token during
   one connection that can be used on a subsequent connection.  Address
   validation is especially important for 0-RTT, where the with 0-RTT because a server
   might wish to provide application data traffic - such as
   potentially sends a response significant amount of data to a request - client in
   response to 0-RTT data.

   A different type of token is needed when resuming.  Unlike the data carried in the early data from
   the client.

   To send additional data prior to completing the cryptographic token
   that is created during a handshake, there might be some time between
   when the server then needs to validate token is created and when the token is subsequently used.

   Thus, a resumption token SHOULD include an expiration time.  It is
   also unlikely that the client owns port number is the
   address that it claims.

   Source address validation same on two
   different connections; validating the port is therefore performed during unlikely to
   be successful.

   This token can be provided to the
   establishment of cryptographic handshake immediately
   after establishing a connection.  TLS provides the tools that support
   the feature, but basic validation is performed by the core transport
   protocol.

7.4.1.  Client Address Validation Procedure  QUIC uses token-based address validation.  Any might also generate an updated
   token if significant time the server wishes
   to validate a client address, it provides passes or the client with a token.
   As long as the token cannot be easily guessed address changes for
   any reason (see Section 7.4.3), if
   the client 7.6).  The cryptographic handshake is able to return that token, it proves to
   responsible for providing the server that
   it received client with the token.

   During  In TLS the processing of
   token is included in the cryptographic handshake messages from a
   client, TLS will request ticket that QUIC make is used for resumption and
   0-RTT, which is carried in a decision about whether NewSessionTicket message.

7.5.3.  Address Validation Token Integrity

   An address validation token MUST be difficult to
   proceed based on guess.  Including a
   large enough random value in the information it has.  TLS will provide QUIC with
   any token that was provided by would be sufficient, but this
   depends on the client.  For an initial packet,
   QUIC can decide to abort server remembering the connection, allow value it sends to proceed, or
   request address validation.

   If QUIC decides to request address validation, it provides clients.

   A token-based scheme allows the
   cryptographic handshake server to offload any state
   associated with a token.  The contents of validation to the client.  For this design to work,
   the token are
   consumed MUST be covered by integrity protection against
   modification or falsification by clients.  Without integrity
   protection, malicious clients could generate or guess values for
   tokens that would be accepted by the server.  Only the server that generates
   requires access to the token, so there is no need integrity protection key for a single well-defined format.  A token could include information
   about tokens.

   In TLS the claimed client address (IP and port), a timestamp, and any
   other supplementary validation token is often bundled with the
   information that TLS requires, such as the server will need resumption secret.  In
   this case, adding integrity protection can be delegated to validate the
   token in the future.

   The
   cryptographic handshake protocol, avoiding redundant protection.  If
   integrity protection is responsible for enacting validation by
   sending the address validation token delegated to the client.  A legitimate
   client cryptographic handshake, an
   integrity failure will include a copy of the token when it attempts to continue
   the handshake.  The result in immediate cryptographic handshake extracts the token then
   asks QUIC a second time whether the token
   failure.  If integrity protection is acceptable.  In
   response, performed by QUIC, QUIC can either MUST
   abort the connection or permit it to
   proceed.

   A connection MAY be accepted without address validation - or if the integrity check fails with
   only limited validation - but a server SHOULD limit the data it sends
   toward an unvalidated address.  Successful completion of the
   cryptographic handshake implicitly provides proof that
   QUIC_ADDRESS_VALIDATION_FAILURE error code.

7.6.  Connection Migration

   QUIC connections are identified by their 64-bit Connection ID.
   QUIC's consistent connection ID allows connections to survive changes
   to the client's IP and/or port, such as those caused by client has
   received packets from the server.

7.4.2.  Address Validation on Session Resumption

   A or
   server MAY provide clients migrating to a new network.  Connection migration allows a
   client to retain any shared state with an address validation token during
   one a connection when they move
   networks.  This includes state that can be used hard to recover such as
   outstanding requests, which might otherwise be lost with no easy way
   to retry them.

7.6.1.  Privacy Implications of Connection Migration

   Using a stable connection ID on multiple network paths allows a subsequent connection.  Address
   validation is especially important with 0-RTT because
   passive observer to correlate activity between those paths.  A client
   that moves between networks might not wish to have their activity
   correlated by any entity other than a server
   potentially sends server.  The NEW_CONNECTION_ID
   message can be sent by a significant amount of data server to a client provide an unlinkable connection
   ID for use in
   response to 0-RTT data.

   A different type of token is needed when resuming.  Unlike case the token
   that is created during a handshake, there might be some time client wishes to explicitly break linkability
   between
   when two points of network attachment.

   A client which wishes to break linkability upon changing networks
   MUST use the token is created and when NEW_CONNECTION_ID as well as incrementing the token is subsequently used.
   Thus, a resumption token SHOULD include packet
   sequence number by an expiration time.  It is
   also unlikely externally unpredictable value computed as
   described in Section 7.6.1.1.  Packet number gaps are cumulative.  A
   client might skip connection IDs, but it MUST ensure that it applies
   the client port associated packet number is the same on two
   different connections; validating the port is therefore unlikely to
   be successful.

   This token can be provided gaps in addition to the cryptographic handshake immediately
   after establishing a connection.  QUIC might also generate an updated
   token if significant time passes or packet number
   gap associated with the connection ID that it does use.

   A client address changes for might need to send packets on multiple networks without
   receiving any reason (see Section 7.5).  The cryptographic handshake is
   responsible for providing the client with the token.  In TLS the
   token is included in response from the ticket server.  To ensure that the client is used for resumption
   not linkable across each of these changes, a new connection ID and
   0-RTT, which is carried in
   packet number gap are needed for each network.  To support this, a NewSessionTicket message.

7.4.3.  Address Validation Token Integrity

   An address validation token
   server sends multiple NEW_CONNECTION_ID messages.  Each
   NEW_CONNECTION_ID is marked with a sequence number.  Connection IDs
   MUST be difficult to guess.  Including a
   large enough random value used in the token would be sufficient, but this
   depends on the order in which they are numbered.

   A server remembering that receives a packet that is marked with a new connection
   ID recovers the value it sends packet number by adding the cumulative packet number
   gap to clients. its expected packet number.  A token-based scheme allows the server to offload any state SHOULD discard packets
   that contain a smaller gap than it advertised.

   For instance, a server might provide a packet number gap of 7
   associated with validation to the client.  For this design to work,
   the token MUST be covered by integrity protection against
   modification or falsification by clients.  Without integrity
   protection, malicious clients could generate or guess values for
   tokens that would be accepted by the server.  Only a new connection ID.  If the server
   requires access to the integrity protection key for tokens.

   In TLS the address validation token is often bundled with received packet
   10 using the
   information that TLS requires, such as previous connection ID, it should expect packets on the resumption secret.  In
   this case, adding integrity protection can be delegated
   new connection ID to start at 18.  A packet with the
   cryptographic handshake protocol, avoiding redundant protection.  If
   integrity protection new connection
   ID and a packet number of 17 is delegated to the cryptographic handshake, an
   integrity failure will result discarded as being in immediate cryptographic handshake
   failure.  If integrity protection is performed by QUIC, QUIC MUST
   abort the connection if error.

7.6.1.1.  Packet Number Gap

   In order to avoid linkage, the integrity check fails with packet number gap MUST be externally
   indistinguishable from random.  The packet number gap for a
   QUIC_ADDRESS_VALIDATION_FAILURE error code.

7.5.  Connection Migration

   QUIC connections are identified by their 64-bit Connection ID.
   QUIC's consistent
   connection ID allows connections to survive changes
   to with sequence number is computed by encoding the client's IP and/or port, such
   sequence number as those caused by client or
   server migrating to a new network.  QUIC also provides automatic
   cryptographic verification 32-bit integer in big-endian format, and then
   computing:

   Gap = HKDF-Expand-Label(packet_number_secret,
                           "QUIC packet sequence gap", sequence, 4)

   The output of HKDF-Expand-Label is interpreted as a client which has changed its IP
   address because the client continues to use big-endian
   number. "packet_number_secret" is derived from the same session TLS key exchange,
   as described in [QUIC-TLS] Section 5.6.

7.6.2.  Address Validation for
   encrypting and decrypting packets.

   DISCUSS: Simultaneous migration.  Is this reasonable? Migrated Connections

   TODO: Perhaps move mitigation techniques from Security Considerations
   here.

7.6. see issue #161

7.7.  Connection Termination

   Connections should remain open until they become idle for a pre-
   negotiated period of time.  A QUIC connection, once established, can
   be terminated in one of three ways:

   1.  Explicit Shutdown: An endpoint sends a CONNECTION_CLOSE frame to
       initiate a connection termination.  An endpoint may send a GOAWAY
       frame to the peer prior to a CONNECTION_CLOSE to indicate that
       the connection will soon be terminated.  A GOAWAY frame signals
       to the peer that any active streams will continue to be
       processed, but the sender of the GOAWAY will not initiate any
       additional streams and will not accept any new incoming streams.
       On termination of the active streams, a CONNECTION_CLOSE may be
       sent.  If an endpoint sends a CONNECTION_CLOSE frame while
       unterminated streams are active (no FIN bit or RST_STREAM frames
       have been sent or received for one or more streams), then the
       peer must assume that the streams were incomplete and were
       abnormally terminated.

   2.  Implicit Shutdown: The default idle timeout for a QUIC connection
       is 30 seconds, and is a required
       parameter in connection negotiation.  The maximum is 10 minutes.
       If there is no network activity for the duration of the idle
       timeout, the connection is closed.  By default a CONNECTION_CLOSE
       frame will be sent.  A silent close option can be enabled when it
       is expensive to send an explicit close, such as mobile networks
       that must wake up the radio.

   3.  Abrupt Shutdown: An endpoint may send a Public Reset packet at
       any time during the connection to abruptly terminate an active
       connection.  A Public Reset packet SHOULD only be used as a final
       recourse.  Commonly, a public reset is expected to be sent when a
       packet on an established connection is received by an endpoint
       that is unable decrypt the packet.  For instance, if a server
       reboots mid-connection and loses any cryptographic state
       associated with open connections, and then receives a packet on
       an open connection, it should send a Public Reset packet in
       return.  (TODO: articulate rules around when a public reset
       should be sent.)

   TODO: Connections that are terminated are added to a TIME_WAIT list
   at the server, so as to absorb any straggler packets in the network.
   Discuss TIME_WAIT list.

8.  Frame Types and Formats

   As described in Section 6, Regular packets contain one or more
   frames.  We now describe the various QUIC frame types that can be
   present in a Regular packet.  The use of these frames and various
   frame header bits are described in subsequent sections.

8.1.  STREAM Frame

   STREAM frames implicitly create a stream and carry stream data.  The
   type byte for a STREAM frame contains embedded flags, and is
   formatted as "1FDOOOSS". "11FDOOSS".  These bits are parsed as follows:

   o  The leftmost bit first two bits must be set to 1, 11, indicating that this is a
      STREAM frame.

   o  "F" is the FIN bit, which is used for stream termination.

   o  The "D" bit indicates whether a Data Length field is present in
      the STREAM header.  When set to 0, this field indicates that the
      Stream Data field extends to the end of the packet.  When set to
      1, this field indicates that Data Length field contains the length
      (in bytes) of the Stream Data field.  The option to omit the
      length should only be used when the packet is a "full-sized"
      packet, to avoid the risk of corruption via padding.

   o  The "OOO" "OO" bits encode the length of the Offset header field as 0,
      16, 24, 32, 40, 48, 56, or 64 bits long.

   o  The "SS" bits encode the length of the Stream ID header field as
      8, 16, 24, or 32 bits.  (DISCUSS: Consider making this 8, 16, 32,
      64.)

   A STREAM frame is shown below.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |       [Data Length (16)]      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Stream ID (8/16/24/32)                   ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      Offset (0/16/24/32/40/48/56/64) (0/16/32/64)                    ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Stream Data (*)                      ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                       Figure 7: STREAM Frame Format

   The STREAM frame contains the following fields:

   Data Length:  An optional 16-bit unsigned number specifying the
      length of the Stream Data field in this STREAM frame.  This field
      is present when the "D" bit is set to 1.

   Stream ID:  A variable-sized unsigned  The stream ID unique to this stream. of the stream (see Section 10.1).

   Offset:  A variable-sized unsigned number specifying the byte offset
      in the stream for the data in this STREAM frame.  When the offset
      length is 0, the offset is 0.  The first byte in the stream has an
      offset of 0.  The largest offset delivered on a stream - the sum
      of the re-constructed offset and data length - MUST be less than
      2^64.

   Stream Data:  The bytes from the designated stream to be delivered.

   A STREAM frame MUST have either non-zero data length or the FIN bit
   set.  When the FIN flag is sent on an empty STREAM frame, the offset
   in the STREAM frame MUST be one greater than the last data byte sent
   on this stream.

   Stream multiplexing is achieved by interleaving STREAM frames from
   multiple streams into one or more QUIC packets.  A single QUIC packet
   MAY bundle STREAM frames from multiple streams.

   Implementation note: One of the benefits of QUIC is avoidance of
   head-of-line blocking across multiple streams.  When a packet loss
   occurs, only streams with data in that packet are blocked waiting for
   a retransmission to be received, while other streams can continue
   making progress.  Note that when data from multiple streams is
   bundled into a single QUIC packet, loss of that packet blocks all
   those streams from making progress.  An implementation is therefore
   advised to bundle as few streams as necessary in outgoing packets
   without losing transmission efficiency to underfilled packets.

8.2.  ACK Frame

   Receivers send ACK frames to inform senders which packets they have
   received and processed, as well as which packets are considered
   missing.  The ACK frame contains between 1 and 256 ACK blocks.  ACK
   blocks are ranges of acknowledged packets.

   To limit ACK blocks to those that have not yet been received by the
   sender, the receiver SHOULD track which ACK frames have been
   acknowledged by its peer.  Once an ACK frame has been acknowledged,
   the packets it acknowledges SHOULD not be acknowledged again.  To
   handle cases where the

   A receiver that is only sending ACK frames, and hence frames will not receive
   acknowledgments for its packets, it packets.  Sending an occasional MAX_DATA or
   MAX_STREAM_DATA frame as data is received will ensure that
   acknowledgements are generated by a peer.  Otherwise, an endpoint MAY
   send a PING frame at most once per RTT to explicitly request solicit an acknowledgment.

   To limit receiver state or the size of ACK frames, a receiver MAY
   limit the number of ACK blocks it sends.  A receiver can do this even
   without receiving acknowledgment of its ACK frames, with the
   knowledge this could cause the sender to unnecessarily retransmit
   some data. sender to unnecessarily retransmit
   some data.  When this is necessary, the receiver SHOULD acknowledge
   newly received packets and stop acknowledging packets received in the
   past.

   Unlike TCP SACKs, QUIC ACK blocks are cumulative and therefore
   irrevocable.  Once a packet has been acknowledged, even if it does
   not appear in a future ACK frame, it is assumed to be acknowledged.

   QUIC ACK frames contain a timestamp section with up to 255
   timestamps.  Timestamps enable better congestion control, but are not
   required for correct loss recovery, and old timestamps are less
   valuable, so it is not guaranteed every timestamp will be received by
   the sender.  A receiver SHOULD send a timestamp exactly once for each
   received packet containing retransmittable frames.  A receiver MAY
   send timestamps for non-retransmittable packets.  A receiver MUST not
   send timestamps in unprotected packets.

   A sender MAY intentionally skip packet numbers to introduce entropy
   into the connection, to avoid opportunistic acknowledgement attacks.
   The sender MUST close the connection if an unsent packet number is
   acknowledged.  The format of the ACK frame is efficient at expressing
   blocks of missing packets; skipping packet numbers between 1 and 255
   effectively provides up to 8 bits of efficient entropy on demand,
   which should be adequate protection against most opportunistic
   acknowledgement attacks.

   The type byte for a ACK frame contains embedded flags, and is
   formatted as "01NULLMM". "101NLLMM".  These bits are parsed as follows:

   o  The first two three bits must be set to 01 101 indicating that this is an
      ACK frame.

   o  The "N" bit indicates whether the frame has more than 1 range of
      acknowledged packets (i.e., whether the ACK Block Section contains
      a Num Blocks field).

   o  The "U" bit is unused and MUST be set to zero.

   o  The two "LL" bits encode the length of the Largest Acknowledged
      field as 1, 2, 4, or 6 bytes long.

   o  The two "MM" bits encode the length of the ACK Block Length fields
      as 1, 2, 4, or 6 bytes long.

   An ACK frame is shown below.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |[Num Blocks(8)]|   NumTS (8)   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                Largest Acknowledged (8/16/32/48)            ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |        ACK Delay (16)         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     ACK Block Section (*)                   ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Timestamp Section (*)                   ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                        Figure 8: ACK Frame Format

   The fields in the ACK frame are as follows:

   Num Blocks (opt):  An optional 8-bit unsigned value specifying the
      number of additional ACK blocks (besides the required First ACK
      Block) in this ACK frame.  Only present if the 'N' flag bit is 1.

   Num Timestamps:  An unsigned 8-bit number specifying the total number
      of <packet number, timestamp> pairs in the Timestamp Section.

   Largest Acknowledged:  A variable-sized unsigned value representing
      the largest packet number the peer is acknowledging in this packet
      (typically the largest that the peer has seen thus far.)

   ACK Delay:  The time from when the largest acknowledged packet, as
      indicated in the Largest Acknowledged field, was received by this
      peer to when this ACK was sent.

   ACK Block Section:  Contains one or more blocks of packet numbers
      which have been successfully received, see Section 8.2.1.

   Timestamp Section:  Contains zero or more timestamps reporting
      transit delay of received packets.  See Section 8.2.2.

8.2.1.  ACK Block Section

   The ACK Block Section contains between one and 256 blocks of packet
   numbers which have been successfully received.  If the Num Blocks
   field is absent, only the First ACK Block length is present in this
   section.  Otherwise, the Num Blocks field indicates how many
   additional blocks follow the First ACK Block Length field.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |              First ACK Block Length (8/16/32/48)            ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  [Gap 1 (8)]  |       [ACK Block 1 Length (8/16/32/48)]     ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  [Gap 2 (8)]  |       [ACK Block 2 Length (8/16/32/48)]     ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  [Gap N (8)]  |       [ACK Block N Length (8/16/32/48)]     ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                        Figure 9: ACK Block Section

   The fields in the ACK Block Section are:

   First ACK Block Length:  An unsigned packet number delta that
      indicates the number of contiguous additional packets being
      acknowledged starting at the Largest Acknowledged.

   Gap To Next Block (opt, repeated):  An unsigned number specifying the
      number of contiguous missing packets from the end of the previous
      ACK block to the start of the next.  Repeated "Num Blocks" times.

   ACK Block Length (opt, repeated):  An unsigned packet number delta
      that indicates the number of contiguous packets being acknowledged
      starting after the end of the previous gap.  Repeated "Num Blocks"
      times.

8.2.2.  Timestamp Section

   The Timestamp Section contains between zero and 255 measurements of
   packet receive times relative to the beginning of the connection.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+
   | [Delta LA (8)]|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    [First Timestamp (32)]                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |[Delta LA 1(8)]| [Time Since Previous 1 (16)]  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |[Delta LA 2(8)]| [Time Since Previous 2 (16)]  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                          ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |[Delta LA N(8)]| [Time Since Previous N (16)]  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                       Figure 10: Timestamp Section

   The fields in the Timestamp Section are:

   Delta Largest Acknowledged (opt):  An optional 8-bit unsigned packet
      number delta specifying Timestamp Section are:

   Delta Largest Acknowledged (opt):  An optional 8-bit unsigned packet
      number delta specifying the delta between the largest acknowledged
      and the first packet whose timestamp is being reported.  In other
      words, this first packet number may be computed as (Largest
      Acknowledged - Delta Largest Acknowledged.)

   First Timestamp (opt):  An optional 32-bit unsigned value specifying
      the time delta in microseconds, from the beginning of the
      connection to the arrival of the packet indicated by Delta Largest
      Acknowledged.

   Delta Largest Acked 1..N (opt, repeated):  This field has the same
      semantics and format as "Delta Largest Acknowledged".  Repeated
      "Num Timestamps - 1" times.

   Time Since Previous Timestamp 1..N(opt, repeated):  An optional
      16-bit unsigned value specifying time delta from the previous
      reported timestamp.  It is encoded in the same format as the ACK
      Delay.  Repeated "Num Timestamps - 1" times.

   The timestamp section lists packet receipt timestamps ordered by
   timestamp.

8.2.2.1.  Time Format

   DISCUSS_AND_REPLACE: Perhaps make this format simpler.

   The time format used in the ACK frame above is a 16-bit unsigned
   float with 11 explicit bits of mantissa and 5 bits of explicit
   exponent, specifying time in microseconds.  The bit format is loosely
   modeled after IEEE 754.  For example, 1 microsecond is represented as
   0x1, which has an exponent of zero, presented in the 5 high order
   bits, and mantissa of 1, presented in the 11 low order bits.  When
   the delta between explicit exponent is greater than zero, an implicit high-order
   12th bit of 1 is assumed in the largest acknowledged mantissa.  For example, a floating
   value of 0x800 has an explicit exponent of 1, as well as an explicit
   mantissa of 0, but then has an effective mantissa of 4096 (12th bit
   is assumed to be 1).  Additionally, the actual exponent is one-less
   than the explicit exponent, and the first value represents 4096
   microseconds.  Any values larger than the representable range are
   clamped to 0xFFFF.

8.2.3.  ACK Frames and Packet Protection

   ACK frames that acknowledge protected packets MUST be carried in a
   packet whose timestamp that has an equivalent or greater level of packet protection.

   Packets that are protected with 1-RTT keys MUST be acknowledged in
   packets that are also protected with 1-RTT keys.

   A packet that is being reported.  In other
      words, this first not protected and claims to acknowledge a packet
   number may be computed as (Largest
      Acknowledged - Delta Largest Acknowledged.)

   First Timestamp (opt): that was sent with packet protection is not valid.  An optional 32-bit unsigned value specifying
   unprotected packet that carries acknowledgments for protected packets
   MUST be discarded in its entirety.

   Packets that a client sends with 0-RTT packet protection MUST be
   acknowledged by the time delta server in microseconds, from packets protected by 1-RTT keys.  This
   can mean that the beginning of client is unable to use these acknowledgments if
   the server cryptographic handshake messages are delayed or lost.
   Note that the
      connection same limitation applies to other data sent by the arrival of the packet indicated
   server protected by Delta Largest
      Acknowledged.

   Delta Largest Acked 1..N (opt, repeated):  This field has the same
      semantics and format 1-RTT keys.

   Unprotected packets, such as "Delta Largest Acknowledged".  Repeated
      "Num Timestamps - 1" times.

   Time Since Previous Timestamp 1..N(opt, repeated):  An optional
      16-bit unsigned value specifying time delta from those that carry the previous
      reported timestamp.  It is encoded initial
   cryptographic handshake messages, MAY be acknowledged in the same format as the ACK
      Delay.  Repeated "Num Timestamps - 1" times.

   The timestamp section lists packet receipt timestamps ordered by
   timestamp.

8.2.2.1.  Time Format

   DISCUSS_AND_REPLACE: Perhaps make this format simpler.

   The time format used unprotected
   packets.  Unprotected packets are vulnerable to falsification or
   modification.  Unprotected packets can be acknowledged along with
   protected packets in the ACK frame above is a 16-bit unsigned
   float with 11 explicit bits of mantissa and 5 bits of explicit
   exponent, specifying time protected packet.

   An endpoint SHOULD acknowledge packets containing cryptographic
   handshake messages in microseconds.  The bit format is loosely
   modeled after IEEE 754.  For example, 1 microsecond the next unprotected packet that it sends,
   unless it is represented as
   0x1, which has an exponent of zero, presented able to acknowledge those packets in later packets
   protected by 1-RTT keys.  At the 5 high order
   bits, and mantissa completion of 1, presented the cryptographic
   handshake, both peers send unprotected packets containing
   cryptographic handshake messages followed by packets protected by
   1-RTT keys.  An endpoint SHOULD acknowledge the unprotected packets
   that complete the cryptographic handshake in a protected packet,
   because its peer is guaranteed to have access to 1-RTT packet
   protection keys.

   For instance, a server acknowledges a TLS ClientHello in the 11 low order bits.  When packet
   that carries the explicit exponent is greater than zero, an implicit high-order
   12th bit of 1 is assumed TLS ServerHello; similarly, a client can acknowledge
   a TLS HelloRetryRequest in the mantissa.  For example, packet containing a floating
   value of 0x800 has an explicit exponent of 1, as well as an explicit
   mantissa of 0, but then has an effective mantissa second TLS
   ClientHello.  The complete set of 4096 (12th bit
   is assumed server handshake messages (TLS
   ServerHello through to Finished) might be 1).  Additionally, acknowledged by a client in
   protected packets, because it is certain that the actual exponent server is one-less
   than able to
   decipher the explicit exponent, and packet.

8.3.  MAX_DATA Frame

   The MAX_DATA frame (type=0x04) is used in flow control to inform the value represents 4096
   microseconds.  Any values larger than
   peer of the representable range are
   clamped to 0xFFFF.

8.2.3.  ACK Frames and Packet Protection

   ACK frames maximum amount of data that acknowledge protected packets MUST can be carried in sent on the connection
   as a
   packet that has an equivalent or greater level whole.

   The frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                        Maximum Data (64)                      +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The fields in the MAX_DATA frame are as follows:

   Maximum Data:  A 64-bit unsigned integer indicating the maximum
      amount of packet protection.

   Packets data that are protected with 1-RTT keys MUST can be acknowledged sent on the entire connection, in
   packets that are also protected with 1-RTT keys.

   A packet that units
      of 1024 octets.  That is, the updated connection-level data limit
      is not protected and claims to acknowledge a packet
   number that was determined by multiplying the encoded value by 1024.

   All data sent in STREAM frames counts toward this limit, with packet protection is not valid. the
   exception of data on stream 0.  The sum of the largest received
   offsets on all streams - including closed streams, but excluding
   stream 0 - MUST NOT exceed the value advertised by a receiver.  An
   unprotected packet that carries acknowledgments for protected packets
   endpoint MUST be discarded in its entirety.

   Packets that terminate a client sends connection with 0-RTT packet protection MUST be
   acknowledged by a
   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA error if it receives more
   data than the server in packets protected by 1-RTT keys.  This
   can mean maximum data value that it has sent, unless this is a
   result of a change in the client initial limits (see Section 7.3.2).

8.4.  MAX_STREAM_DATA Frame

   The MAX_STREAM_DATA frame (type=0x05) is unable used in flow control to use these acknowledgments if
   inform a peer of the maximum amount of data that can be sent on a
   stream.

   The frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Stream ID (32)                         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                    Maximum Stream Data (64)                   +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The fields in the server cryptographic handshake messages MAX_STREAM_DATA frame are delayed or lost.
   Note as follows:

   Stream ID:  The stream ID of the stream that is affected.

   Maximum Stream Data:  A 64-bit unsigned integer indicating the same limitation applies to other
      maximum amount of data sent by the
   server protected by the 1-RTT keys.

   Unprotected packets, such as those that carry the initial
   cryptographic handshake messages, MAY be acknowledged in unprotected
   packets.  Unprotected packets are vulnerable to falsification or
   modification.  Unprotected packets can be acknowledged along with
   protected packets in a protected packet.

   An endpoint SHOULD acknowledge packets containing cryptographic
   handshake messages in sent on the next unprotected packet that it sends,
   unless it is able to acknowledge those packets identified stream,
      in later packets
   protected by 1-RTT keys.  At the completion units of the cryptographic
   handshake, both peers send unprotected packets containing
   cryptographic handshake messages followed by packets protected by
   1-RTT keys.  An octets.

   When counting data toward this limit, an endpoint SHOULD acknowledge accounts for the unprotected packets
   largest received offset of data that complete the cryptographic handshake in a protected packet,
   because its peer is guaranteed to have access to 1-RTT packet
   protection keys.

   For instance, a server acknowledges a TLS ClientHello in sent or received on the packet
   stream.  Loss or reordering can mean that carries the TLS ServerHello; similarly, largest received offset
   on a client stream can acknowledge be greater than the total size of data received on
   that stream.  Receiving STREAM frames might not increase the largest
   received offset.

   The data sent on a TLS HelloRetryRequest in stream MUST NOT exceed the largest maximum stream
   data value advertised by the packet containing receiver.  An endpoint MUST terminate a second TLS
   ClientHello.  The complete set of server handshake messages (TLS
   ServerHello through to Finished) might be acknowledged by
   connection with a client in
   protected packets, because QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA error if
   it is certain receives more data than the largest maximum stream data that it
   has sent for the server affected stream, unless this is able to
   decipher a result of a change
   in the packet.

8.3.  WINDOW_UPDATE initial limits (see Section 7.3.2).

8.5.  MAX_STREAM_ID Frame

   The WINDOW_UPDATE MAX_STREAM_ID frame (type=0x04) (type=0x06) informs the peer of an increase
   in an endpoint's flow control receive window for either a single
   stream, or the entire connection as a whole. maximum
   stream ID that they are permitted to open.

   The frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Maximum Stream ID (32)                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                    Flow Control Offset (64)                   +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The fields in the WINDOW_UPDATE MAX_STREAM_ID frame are as follows:

   Maximum Stream ID:  ID of the stream whose flow control windows is being
      updated, or 0 to specify the connection-level flow control window.

   Flow Control Offset:  A 64-bit unsigned integer indicating the flow
      control offset for the given stream (for a maximum peer-initiated stream ID other than 0)
      or the entire connection.

   The flow control offset is expressed in units of octets for
   individual streams (for stream identifiers other than 0).

   The connection-level flow control offset is expressed in units of
   1024 octets (for a stream identifier of 0).  That is, the connection-
   level flow control offset is determined by multiplying the encoded
   value by 1024.

   An endpoint accounts for
      the maximum offset of data that is sent or
   received on a stream. connection.

   Loss or reordering can mean that the maximum
   offset is greater than the total size of data a MAX_STREAM_ID frame can be
   received on which states a stream.
   Similarly, receiving STREAM lower stream limit than the client has
   previously received.  MAX_STREAM_ID frames might which do not increase the
   maximum
   offset on a stream.  A STREAM frame with a FIN bit set or RST_STREAM
   causes the final offset for a stream to ID MUST be fixed.

   The maximum data offset on a stream ignored.

   A peer MUST NOT exceed the initiate a stream flow
   control offset advertised by the receiver.  The sum of with a higher stream ID than the
   greatest maximum
   data offsets of all streams (including closed streams) MUST NOT
   exceed the connection flow control offset advertised by the receiver. stream ID it has received.  An endpoint MUST
   terminate a connection with a
   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA QUIC_TOO_MANY_OPEN_STREAMS error if it receives more
   data a
   peer initiates a stream with a higher stream ID than the largest flow control offset that it has sent,
   unless this is a result of a change in the initial offsets limits (see
   Section 7.3.2).

8.4.

8.6.  BLOCKED Frame

   A sender sends a BLOCKED frame (type=0x05) (type=0x08) when it is ready wishes to send
   data (and has data to send),
   data, but is currently unable to due to connection-level flow control blocked. (see
   Section 11.2.1).  BLOCKED frames are purely informational frames, but extremely useful
   for debugging purposes.  A receiver can be used as input to tuning of a
   flow control algorithms (see Section 11.1.2).

   The BLOCKED frame should simply
   discard does not contain a payload.

8.7.  STREAM_BLOCKED Frame

   A sender sends a STREAM_BLOCKED frame (type=0x09) when it (after possibly printing a helpful log message). wishes to
   send data, but is unable to due to stream-level flow control.  This
   frame is analogous to BLOCKED (Section 8.6).

   The STREAM_BLOCKED frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Stream ID (32)                         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The BLOCKED STREAM_BLOCKED frame contains a single field:

   Stream ID:  A 32-bit unsigned number indicating the stream which is
      flow control blocked.  A non-zero Stream ID field specifies the

   An endpoint MAY send a STREAM_BLOCKED frame for a stream that is flow control blocked.  When zero, exceeds
   the Stream maximum stream ID
      field indicates set by its peer (see Section 8.5).  This does
   not open the stream, but informs the peer that a new stream was
   needed, but the connection stream limit prevented the creation of the stream.

8.8.  STREAM_ID_NEEDED Frame

   A sender sends a STREAM_ID_NEEDED frame (type=0x0a) when it wishes to
   open a stream, but is flow control blocked.

8.5. unable to due to the maximum stream ID limit.

   The STREAM_ID_NEEDED frame does not contain a payload.

8.9.  RST_STREAM Frame

   An endpoint may use a RST_STREAM frame (type=0x01) to abruptly
   terminate a stream.

   After sending a RST_STREAM, an endpoint ceases transmission of STREAM
   frames on the identified stream.  A receiver of RST_STREAM can
   discard any data that it already received on that stream.  An
   endpoint sends a RST_STREAM in response to a RST_STREAM unless the
   stream is already closed.

   The RST_STREAM frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Error Code (32)                        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Stream ID (32)                         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                       Final Offset (64)                       +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   The fields are:

   Error code:  A 32-bit error code which indicates why the stream is
      being closed.

   Stream ID:  The 32-bit Stream ID of the stream being terminated.

   Final offset:  A 64-bit unsigned integer indicating the absolute byte
      offset of the end of data written on this stream by the RST_STREAM
      sender.

8.6.

8.10.  PADDING Frame

   The PADDING frame (type=0x00) has no semantic value.  PADDING frames
   can be used to increase the size of a packet.  Padding can be used to
   increase an initial client packet to the minimum required size, or to
   provide protection against traffic analysis for protected packets.

   A PADDING frame has no content.  That is, a PADDING frame consists of
   the single octet that identifies the frame as a PADDING frame.

8.7.

8.11.  PING frame

   Endpoints can use PING frames (type=0x07) to verify that their peers
   are still alive or to check reachability to the peer.  The PING frame
   contains no additional fields.  The receiver of a PING frame simply
   needs to acknowledge the packet containing this frame.  The PING
   frame SHOULD be used to keep a connection alive when a stream is
   open.  The default is to send a PING frame after 15 seconds of
   quiescence.  A PING frame has no additional fields.

8.8.

8.12.  NEW_CONNECTION_ID Frame

   A server sends a NEW_CONNECTION_ID to provide the client with
   alternative connection IDs that can be used to break linkability when
   migrating connections (see Section 7.6.1).

   The NEW_CONNECTION_ID is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |       Sequence (16)           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   +                        Connection ID (64)                     +
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   The fields are:

   Sequence:  A 16-bit sequence number.  This value starts at 0 and
      increases by 1 for each connection ID that is provided by the
      server.  The sequence value can wrap; the value 65535 is followed
      by 0.  When wrapping the sequence field, the server MUST ensure
      that a value with the same sequence has been received and
      acknowledged by the client.  The connection ID that is assigned
      during the handshake is assumed to have a sequence of 65535.

   Connection ID:  A 64-bit connection ID.

8.13.  CONNECTION_CLOSE frame

   An endpoint sends a CONNECTION_CLOSE frame (type=0x02) to notify its
   peer that the connection is being closed.  If there are open streams
   that haven't been explicitly closed, they are implicitly closed when
   the connection is closed.  (Ideally, a GOAWAY frame would be sent
   with enough time that all streams are torn down.)  The frame is as
   follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Error Code (32)                        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Reason Phrase Length (16)   |      [Reason Phrase (*)]    ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The fields of a CONNECTION_CLOSE frame are as follows:

   Error Code:  A 32-bit error code which indicates the reason for
      closing this connection.

   Reason Phrase Length:  A 16-bit unsigned number specifying the length
      of the reason phrase. reason phrase.  Note that a CONNECTION_CLOSE frame cannot
      be split between packets, so in practice any limits on packet size
      will also limit the space available for a reason phrase.

   Reason Phrase:  A human-readable explanation for why the connection
      was closed.  This may can be zero length if the sender chooses to not
      give details beyond the Error Code.

   Reason Phrase:  An optional human-readable explanation for why the
      connection was closed.

8.9.  This SHOULD be a UTF-8
      encoded string [RFC3629].

8.14.  GOAWAY Frame

   An endpoint uses a GOAWAY frame (type=0x03) to initiate a graceful
   shutdown of a connection.  The endpoints will continue to use any
   active streams, but the sender of the GOAWAY will not initiate or
   accept any additional streams beyond those indicated.  The GOAWAY
   frame is as follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                  Largest Client Stream ID (32)                |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                  Largest Server Stream ID (32)                |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The fields of a GOAWAY frame are:

   Largest Client Stream ID:  The highest-numbered, client-initiated
      stream on which the endpoint sending the GOAWAY frame either sent
      data, or received and delivered data.  All higher-numbered,
      client-initiated streams (that is, odd-numbered streams) are
      implicitly reset by sending or receiving the GOAWAY frame.

   Largest Server Stream ID:  The highest-numbered, server-initiated
      stream on which the endpoint sending the GOAWAY frame either sent
      data, or received and delivered data.  All higher-numbered,
      server-initiated streams (that is, even-numbered streams) are
      implicitly reset by sending or receiving the GOAWAY frame.

   A GOAWAY frame indicates that any application layer actions on
   streams with higher numbers than those indicated can be safely
   retried because no data was exchanged.  An endpoint MUST set the
   value of the Largest Client or Server Stream ID to be at least as
   high as the highest-numbered stream on which it either sent data or
   received and delivered data to the application protocol that uses
   QUIC.

   An endpoint MAY choose a larger stream identifier if it wishes to
   allow for a number of streams to be created.  This is especially
   valuable for peer-initiated streams where packets creating new
   streams could be in transit; using a larger stream number allows
   those streams to complete.

   In addition to initiating a graceful shutdown of a connection, GOAWAY
   MAY be sent immediately prior to sending a CONNECTION_CLOSE frame
   that is sent as a result of detecting a fatal error.  Higher-numbered
   streams than those indicated in the GOAWAY frame can then be retried.

9.  Packetization and Reliability

   The Path Maximum Transmission Unit (PTMU) is the maximum size of the
   entire IP header, UDP header, and UDP payload.  The UDP payload
   includes the QUIC public header, encrypted protected payload, and any
   authentication fields.

   All QUIC packets SHOULD be sized to fit within the estimated PMTU to
   avoid IP fragmentation or packet drops.  To optimize bandwidth
   efficiency, endpoints SHOULD use Packetization Layer PMTU Discovery
   ([RFC4821]) and MAY use PMTU Discovery ([RFC1191], [RFC1981]) for
   detecting the PMTU, setting the PMTU appropriately, and storing the
   result of previous PMTU determinations.

   In the absence of these mechanisms, QUIC endpoints SHOULD NOT send IP
   packets larger than 1280 octets.  Assuming the minimum IP header
   size, this results in a UDP payload length QUIC packet size of 1232 octets for IPv6 and
   1252 octets for IPv4.

   QUIC endpoints that implement any kind of PMTU discovery SHOULD
   maintain an estimate for each combination of local and remote IP
   addresses (as each pairing could have a different maximum MTU in the
   path). (as each pairing could have a different maximum MTU in the
   path).

   QUIC depends on the network path supporting a MTU of at least 1280
   octets.  This is the IPv6 minimum and therefore also supported by
   most modern IPv4 networks.  An endpoint MUST NOT reduce their MTU
   below this number, even if it receives signals that indicate a
   smaller limit might exist.

   Clients MUST ensure that the first packet in a connection, and any
   retransmissions of those octets, has a QUIC packet size of least 1232
   octets for an IPv6 packet and 1252 octets for an IPv4 packet.  In the
   absence of extensions to the IP header, padding to exactly these
   values will result in an IP packet that is 1280 octets.

   The initial client packet SHOULD be padded to exactly these values
   unless the client has a reasonable assurance that the PMTU is larger.
   Sending a packet of this size ensures that the network path supports
   an MTU of this size and helps reduce the amplitude of amplification
   attacks caused by server responses toward an unverified client
   address.

   Servers MUST ignore an initial plaintext packet from a client if its
   total size is less than 1232 octets for IPv6 or 1252 octets for IPv4.

   If a QUIC endpoint determines that the PMTU between any pair of local
   and remote IP addresses has fallen below 1280 octets, it MUST
   immediately cease sending QUIC depends packets on the network path supporting a MTU of at least 1280
   octets. affected path.  This is
   could result in termination of the IPv6 minimum and therefore also supported by
   most modern IPv4 networks.  An endpoint MUST NOT reduce their MTU
   below this number, even connection if it receives signals that indicate a
   smaller limit might exist.

   Clients MUST ensure that the first packet an alternative path
   cannot be found.

   A sender bundles one or more frames in a connection, Regular QUIC packet (see
   Section 6).

   A sender SHOULD minimize per-packet bandwidth and any
   retransmissions computational costs
   by bundling as many frames as possible within a QUIC packet.  A
   sender MAY wait for a short period of those octets, has time to bundle multiple frames
   before sending a total size (including IP and
   UDP headers) packet that is not maximally packed, to avoid
   sending out large numbers of at least 1280 bytes. small packets.  An implementation may
   use heuristics about expected application sending behavior to
   determine whether and for how long to wait.  This might require inclusion waiting period is
   an implementation decision, and an implementation should be careful
   to delay conservatively, since any delay is likely to increase
   application-visible latency.

   Regular QUIC packets are "containers" of
   PADDING frames.  It frames; a packet is RECOMMENDED that never
   retransmitted whole.  How an endpoint handles the loss of the frame
   depends on the type of the frame.  Some frames are simply
   retransmitted, some have their contents moved to new frames, and
   others are never retransmitted.

   When a packet is detected as lost, the sender re-sends any frames as
   necessary:

   o  All application data sent in STREAM frames MUST be padded to exactly
   1280 octets retransmitted,
      unless the client endpoint has sent a reasonable assurance RST_STREAM for that the
   PMTU is larger.  Sending stream.  When
      an endpoint sends a packet of this size ensures RST_STREAM frame, data outstanding on that
      stream SHOULD NOT be retransmitted, since subsequent data on this
      stream is expected to not be delivered by the
   network path supports an MTU receiver.

   o  ACK and PADDING frames MUST NOT be retransmitted.  ACK frames are
      cumulative, so new frames containing updated information will be
      sent as described in Section 8.2.

   o  All other frames MUST be retransmitted.

   Upon detecting losses, a sender MUST take appropriate congestion
   control action.  The details of this size loss detection and helps mitigate
   amplification attacks caused by server responses toward an unverified
   client address.

   Servers congestion control
   are described in [QUIC-RECOVERY].

   A packet MUST reject NOT be acknowledged until packet protection has been
   successfully removed and all frames contained in the first plaintext packet received from a client
   if it its total size is less than 1280 octets, to mitigate
   amplification attacks.

   If a QUIC endpoint determines that have been
   processed.  For STREAM frames, this means the PMTU between any pair of local
   and remote IP addresses data has fallen below 1280 octets, it MUST
   immediately cease sending QUIC packets between those IP addresses. been queued
   (but not necessarily delivered to the application).  This may result in abrupt termination of also means
   that any stream state transitions triggered by STREAM or RST_STREAM
   frames have occurred.  Once the connection if all pairs
   are affected.  In this case, an endpoint SHOULD send a Public Reset packet to indicate the failure.  The application SHOULD attempt to
   use TLS over TCP instead.

   A sender bundles has been fully processed, a
   receiver acknowledges receipt by sending one or more ACK frames in a Regular QUIC
   containing the packet (see
   Section 6).

   A sender SHOULD minimize per-packet bandwidth and computational costs
   by bundling as many frames as possible within a QUIC packet.  A
   sender MAY wait for a short period number of time to bundle multiple frames
   before sending a packet that is not maximally packed, to the received packet.

   To avoid
   sending out large numbers of small packets.  An implementation may
   use heuristics about expected application sending behavior to
   determine whether and for how long to wait.  This waiting period is creating an implementation decision, and indefinite feedback loop, an implementation should be careful
   to delay conservatively, since any delay is likely endpoint MUST NOT
   generate an ACK frame in response to increase
   application-visible latency.

   Regular QUIC packets are "containers" of frames; a packet is never
   retransmitted whole.  How an endpoint handles containing only ACK or
   PADDING frames.

   Strategies and implications of the loss frequency of generating
   acknowledgments are discussed in more detail in [QUIC-RECOVERY].

9.1.  Special Considerations for PMTU Discovery

   Traditional ICMP-based path MTU discovery in IPv4 [RFC1191] is
   potentially vulnerable to off-path attacks that successfully guess
   the frame
   depends on IP/port 4-tuple and reduce the type of MTU to a bandwidth-inefficient
   value.  TCP connections mitigate this risk by using the frame.  Some frames are simply
   retransmitted, some have their contents moved (at minimum)
   8 bytes of transport header echoed in the ICMP message to new frames, and
   others are never retransmitted.

   When a packet is detected validate
   the TCP sequence number as lost, valid for the sender re-sends any frames current connection.
   However, as
   necessary:

   o  All application data sent QUIC operates over UDP, in STREAM frames MUST be retransmitted,
      unless IPv4 the endpoint echoed information
   could consist only of the IP and UDP headers, which usually has sent a RST_STREAM for that stream.  When
      an endpoint sends
   insufficient entropy to mitigate off-path attacks.

   As a RST_STREAM frame, data outstanding on result, endpoints that
      stream implement PMTUD in IPv4 SHOULD NOT be retransmitted, since subsequent data on this
      stream is expected take steps
   to not be delivered by the receiver. mitigate this risk.  For instance, an application could:

   o  ACK and PADDING frames MUST NOT be retransmitted.  ACK frames are
      cumulative,  Set the IPv4 Don't Fragment (DF) bit on a small proportion of
      packets, so new frames containing updated information will that most invalid ICMP messages arrive when there are
      no DF packets outstanding, and can therefore be
      sent identified as described in Section 8.2.
      spurious.

   o  All other frames MUST be retransmitted.

   Upon detecting losses,  Store additional information from the IP or UDP headers from DF
      packets (for example, the IP ID or UDP checksum) to further
      authenticate incoming Datagram Too Big messages.

   o  Any reduction in PMTU due to a sender MUST take appropriate congestion
   control action.  The details of loss detection and congestion control
   are described report contained in [QUIC-RECOVERY].

   A an ICMP packet MUST NOT be acknowledged
      is provisional until QUIC's loss detection algorithm determines
      that the packet protection has been
   successfully removed and all frames contained is actually lost.

10.  Streams: QUIC's Data Structuring Abstraction

   Streams in QUIC provide a lightweight, ordered, and bidirectional
   byte-stream abstraction modeled closely on HTTP/2 streams [RFC7540].

   Streams can be created either by the packet have been
   processed.  For STREAM frames, this means client or the server, can
   concurrently send data has been queued
   (but not necessarily interleaved with other streams, and can be
   cancelled.

   Data that is received on a stream is delivered in order within that
   stream, but there is no particular delivery order across streams.
   Transmit ordering among streams is left to the application).  This also means
   that any stream state transitions triggered by STREAM or RST_STREAM
   frames implementation.

   The creation and destruction of streams are expected to have occurred.  Once the packet has been fully processed, minimal
   bandwidth and computational cost.  A single STREAM frame may create,
   carry data for, and terminate a
   receiver acknowledges receipt by sending one stream, or more ACK frames
   containing a stream may last the packet number
   entire duration of the received packet.

   To avoid creating an indefinite feedback loop, a connection.

   Streams are individually flow controlled, allowing an endpoint MUST NOT
   generate an ACK frame in response to a packet containing only ACK or
   PADDING frames.

   Strategies
   limit memory commitment and implications to apply back pressure.  The creation of
   streams is also flow controlled, with each peer declaring the frequency of generating
   acknowledgments are discussed in more detail in [QUIC-RECOVERY].

9.1.  Special Considerations for PMTU Discovery

   Traditional ICMP-based path MTU discovery in IPv4 ([RFC1191] maximum
   stream ID it is willing to accept at a given time.

   An alternative view of QUIC streams is
   potentially vulnerable as an elastic "message"
   abstraction, similar to off-path attacks that successfully guess
   the IP/port 4-tuple and reduce the MTU to way ephemeral streams are used in SST
   [SST], which may be a bandwidth-inefficient
   value.  TCP connections mitigate this risk more appealing description for some
   applications.

10.1.  Stream Identifiers

   Streams are identified by using an unsigned 32-bit integer, referred to as
   the (at minimum)
   8 bytes of transport header echoed in Stream ID.  To avoid Stream ID collision, clients initiate
   streams using odd-numbered Stream IDs; streams initiated by the ICMP message to validate
   server use even-numbered Stream IDs.

   Stream ID 0 (0x0) is reserved for the TCP sequence number as valid cryptographic handshake.
   Stream 0 MUST NOT be used for application data, and is the current connection.
   However, as first
   client-initiated stream.

   A QUIC operates over UDP, endpoint cannot reuse a Stream ID.  Streams MUST be created in IPv4 the echoed information
   could consist only
   sequential order.  Open streams can be used in any order.  Streams
   that are used out of order result in lower-numbered streams in the IP and UDP headers, which
   same direction being counted as open.

   Stream IDs are usually has
   insufficient entropy to mitigate off-path attacks.

   As encoded as a result, endpoints that implement PMTUD in IPv4 SHOULD take steps
   to mitigate this risk.  For instance, an application could:

   o  Set 32-bit integer, though the IPv4 Don't Fragment (DF) bit on STREAM
   frame (Section 8.1) permits a small proportion of
      packets, so that most invalid ICMP messages arrive when there are
      no DF packets outstanding, and can therefore be identified as
      spurious.

   o  Store additional information from shorter encoding when the IP or UDP headers from DF
      packets (for example, leading bits
   of the IP stream ID or UDP checksum) are zero.

10.2.  Life of a Stream

   The semantics of QUIC streams is based on HTTP/2 streams, and the
   lifecycle of a QUIC stream therefore closely follows that of an
   HTTP/2 stream [RFC7540], with some differences to further
      authenticate incoming Datagram Too Big messages.

   o  Any reduction in PMTU accommodate the
   possibility of out-of-order delivery due to a report contained the use of multiple
   streams in an ICMP packet QUIC.  The lifecycle of a QUIC stream is provisional until QUIC's loss detection algorithm determines
      that shown in the packet is actually lost.

10.  Streams: QUIC's Data Structuring Abstraction

   Streams
   following figure and described below.

                               +--------+
                    recv RST   |        |    send RST
                 ,-------------|  idle  |---------------.
                /              |        |                \
               |               +--------+                 |
               |                    |                     |
               |        send STREAM / recv STREAM         |
               |                    |                     |
               |                    v                     |
               |    recv FIN/  +--------+    send FIN/    |
               |    recv RST   |        |    send RST     |
               |     ,---------|  open  |-----------.     |
               |    /          |        |            \    |
               v   v           +--------+             v   v
            +----------+                          +----------+
            |   half   |                          |   half   |
            |  closed  |                          |  closed  |
            | (remote) |                          |  (local) |
            +----------+                          +----------+
                |                                        |
                |   send FIN/  +--------+    recv FIN/   |
                 \  send RST   |        |    recv RST   /
                  `----------->| closed |<-------------'
                               |        |
                               +--------+

      send:   endpoint sends this frame
      recv:   endpoint receives this frame

      STREAM: a STREAM frame
      FIN:    FIN flag in QUIC provide a lightweight, ordered, STREAM frame
      RST:    RST_STREAM frame

                     Figure 11: Lifecycle of a stream

   Note that this diagram shows stream state transitions and bidirectional
   byte-stream abstraction modeled closely on HTTP/2 streams [RFC7540].

   Streams can be created either by the client or the server, can
   concurrently send data interleaved with other streams, frames
   and can be
   cancelled.

   Data flags that affect those transitions only.  For the purpose of
   state transitions, the FIN flag is received on processed as a stream is delivered in order within that
   stream, but there is no particular delivery order across streams.
   Transmit ordering among streams is left separate event to
   the implementation. frame that bears it; a STREAM frame with the FIN flag set can
   cause two state transitions.

   The creation and destruction recipient of streams are expected to have minimal
   bandwidth and computational cost.  A single STREAM a frame may create,
   carry data for, and terminate which changes stream state will have a stream, or
   delayed view of the state of a stream may last while the
   entire duration frame is in transit.
   Endpoints do not coordinate the creation of a connection.

   Streams streams; they are individually flow controlled, allowing an endpoint to
   limit memory commitment and created
   unilaterally by either endpoint.  Endpoints can use acknowledgments
   to apply back pressure.

   An alternative understand the peer's subjective view of QUIC stream state at any given
   time.

   In the absence of more specific guidance elsewhere in this document,
   implementations SHOULD treat the receipt of a frame that is not
   expressly permitted in the description of a state as a connection
   error (see Section 12).

10.2.1.  idle

   All streams start in the "idle" state.

   The following transitions are valid from this state:

   Sending or receiving a STREAM frame causes the identified stream to
   become "open".  The stream identifier for a new stream is selected as an elastic "message"
   abstraction, similar to the way ephemeral streams are used
   described in SST
   [SST], which may be a more appealing description for some
   applications. Section 10.1.  Life of a Stream  The semantics of QUIC streams same STREAM frame can also cause a
   stream to immediately become "half-closed" if the FIN flag is based set.

   Receiving a STREAM frame on HTTP/2 streams, and the
   lifecycle of a QUIC peer-initiated stream therefore closely follows that of (that is, a
   packet sent by a server on an
   HTTP/2 even-numbered stream [RFC7540], with some differences to accommodate or a client packet
   on an odd-numbered stream) also causes all lower-numbered "idle"
   streams in the
   possibility of out-of-order delivery due same direction to the use of multiple become "open".  This could occur if
   a peer begins sending on streams in QUIC.  The lifecycle of a QUIC stream is shown in the
   following figure and described below.

                               +--------+
                               |        |
                               |  idle  |
                               |        |
                               +--------+
                                    |
                                    | send data/
                                    | recv data/
                                    | recv higher stream
                                    |
                                    v
                               +--------+
                   recv FIN    |        |    send FIN
                     ,---------|  open  |-----------.
                    /          |        |            \
                   v           +--------+             v
            +----------+            |             +----------+
            |   half   |            |             |   half   |
            |  closed  |            | send RST/   |  closed  |
            | (remote) |            | recv RST    |  (local) |
            +----------+            |             +----------+
                |                   |                    |
                | send FIN/         |          recv FIN/ |
                | send RST/         v          send RST/ |
                | recv RST     +--------+      recv RST  |
                `------------->|        |<---------------'
                               | closed |
                               |        |
                               +--------+

      send:   endpoint sends this different order to their
   creation, or it could happen if packets are lost or reordered in
   transit.

   A RST_STREAM frame
      recv: on an "idle" stream causes the stream to become
   "half-closed".  Sending a RST_STREAM frame causes the stream to
   become "half-closed (local)"; receiving RST_STREAM causes the stream
   to become "half-closed (remote)".

   An endpoint receives this might receive MAX_STREAM_DATA frames on peer-initiated
   streams that are "idle" if there is loss or reordering of packets.

   Receiving any frame

      data: application data in other than STREAM, MAX_STREAM_DATA,
   STREAM_BLOCKED, or RST_STREAM on a STREAM frame
      FIN: FIN flag stream in this state MUST be
   treated as a connection error (Section 12) of type YYYY.

   An endpoint MUST NOT send a STREAM frame
      RST: or RST_STREAM frame

                     Figure 11: Lifecycle of for a stream

   Note
   ID that this diagram shows is higher than the peers advertised maximum stream state transitions and ID (see
   Section 8.5).

10.2.2.  open

   A stream in the "open" state may be used by both peers to send frames
   of any type.  In this state, endpoints can send MAX_STREAM_DATA and flags that affect those transitions only.  For
   MUST observe the purpose value advertised by its receiving peer (see
   Section 11).

   From this state, either endpoint can send a frame with the FIN flag
   set, which causes the stream to transition into one of
   state transitions, the "half-
   closed" states.  An endpoint sending an FIN flag is processed as causes the stream
   state to become "half-closed (local)".  An endpoint receiving a separate event FIN
   flag causes the stream state to become "half-closed (remote)" once
   all preceding data has arrived.  The receiving endpoint MUST NOT
   consider the stream state to have changed until all data has arrived.

   A RST_STREAM frame on an "open" stream causes the stream to become
   "half-closed".  Sending a RST_STREAM frame causes the stream to
   become "half-closed (local)"; receiving RST_STREAM causes the stream
   to become "half-closed (remote)".

   Any frame type that bears it; mentions a STREAM frame with the FIN flag set stream ID can
   cause two state transitions.  When the FIN flag is be sent on an empty
   STREAM frame, the offset in this state.

10.2.3.  half-closed (local)

   A stream that is in the STREAM frame "half-closed (local)" state MUST NOT be one greater than
   the last used
   for sending on new STREAM frames.  Retransmission of data byte that has
   already been sent on STREAM frames is permitted.  An endpoint MAY
   also send MAX_STREAM_DATA and RST_STREAM in this stream.

   The recipient of state.

   An endpoint that closes a frame which changes stream state will have a
   delayed view of MUST NOT send data beyond the final
   offset that it has chosen, see Section 10.2.5 for details.

   A stream transitions from this state of to "closed" when a stream while the STREAM frame is in transit.
   Endpoints do not coordinate the creation of streams; they are created
   unilaterally by either endpoint.  The negative consequences of
   that contains a
   mismatch in states are limited to the "closed" state after sending
   RST_STREAM, where frames might be FIN flag is received for some time after
   closing.  Endpoints and all prior data has arrived,
   or when a RST_STREAM frame is received.

   An endpoint can use acknowledgments to understand the peer's
   subjective view of stream state at receive any given time.

   Streams have the following states:

10.1.1.  idle

   All streams start frame that mentions a stream ID in the "idle" state.

   The following transitions are valid from this state:

   Sending or
   state.  Providing flow-control credit using MAX_STREAM_DATA frames is
   necessary to continue receiving flow-controlled frames.  In this
   state, a receiver MAY ignore MAX_STREAM_DATA frames for this stream,
   which might arrive for a short period after a STREAM frame causes bearing the FIN
   flag is sent.

10.2.4.  half-closed (remote)

   A stream to become
   "open".  The stream identifier is selected as described in
   Section 10.2.  The same STREAM frame can also cause a "half-closed (remote)" when the stream is no longer being
   used by the peer to
   immediately become "half-closed".

   Receiving send any data.  An endpoint will have either
   received all data that a STREAM peer has sent or will have received a
   RST_STREAM frame on and discarded any received data.

   Once all data has been either received or discarded, a peer-initiated stream (that is, sender is no
   longer obligated to update the maximum received data for the
   connection.

   An endpoint that receives a
   packet RST_STREAM frame (and which has not sent by
   a server on an even-numbered stream FIN or a client packet RST_STREAM) MUST immediately respond with a RST_STREAM
   frame, and MUST NOT send any more data on an odd-numbered stream) also causes all lower-numbered "idle"
   streams in the same direction stream.

   Due to become "open".  This reordering, an endpoint could occur if continue receiving frames for
   the stream even after the stream is closed for sending.  Frames
   received after a peer begins sending on streams in closes a different order stream SHOULD be discarded.  An
   endpoint MAY choose to their
   creation, or limit the period over which it could happen if packets are lost or reordered in
   transit.

   Receiving any frame other than STREAM or RST_STREAM on a stream in ignores frames
   and treat frames that arrive after this state MUST be treated time as being in error.

   An endpoint will know the final offset of the data it receives on a connection error (Section 12) of type
   YYYY.

10.1.2.  open
   stream when it reaches the "half-closed (remote)" state, see
   Section 11.3 for details.

   A stream in the "open" this state may can be used by both peers the endpoint to send frames
   of any type. frame
   that mentions a stream ID.  In this state, a sending peer must observe the flow-
   control limit endpoint MUST observe
   advertised by its receiving peer (Section stream and connection data limits (see Section 11).

   From this state, either endpoint can send a frame with the FIN flag
   set, which causes the

   A stream to can transition into one from this state to "closed" by completing
   transmission of the "half-
   closed" states.  An endpoint all data.  This includes sending an FIN flag causes all data carried in
   STREAM frames up including the stream
   state to become "half-closed (local)".  An endpoint receiving terminal STREAM frame that contains a
   FIN flag causes the stream state to become "half-closed (remote)" once
   all preceding data has arrived.  The and receiving endpoint MUST NOT
   consider acknowledgment from the stream state to have changed until peer for all data has arrived.

   Either data.

   A stream becomes "closed" when the endpoint can send sends and receives
   acknowledgment of a RST_STREAM frame from this state, causing
   it to transition immediately to "closed".

10.1.3.  half-closed (local)

   A stream that frame.

10.2.5.  closed

   The "closed" state is in the "half-closed (local)" terminal state MUST NOT be used for sending STREAM frames; WINDOW_UPDATE and RST_STREAM MAY a stream.

   Once a stream reaches this state, no frames can be sent that mention
   the stream.  Reordering might cause frames to be received after
   closing, see Section 10.2.4.

10.3.  Stream Concurrency

   An endpoint limits the number of concurrently active incoming streams
   by adjusting the maximum stream ID.  An initial value is set in this state.

   A the
   transport parameters (see Section 7.3.1) and is subsequently
   increased by MAX_STREAM_ID frames (see Section 8.5).

   The maximum stream transitions from this state to "closed" when a STREAM frame
   that contains a FIN flag ID is received specific to each endpoint and all prior data has arrived,
   or when either applies only
   to the peer sends a RST_STREAM frame.

   An endpoint that closes a receives the setting.  That is, clients specify the
   maximum stream ID the server can initiate, and servers specify the
   maximum stream ID the client can initiate.  Each endpoint may respond
   on streams initiated by the other peer, regardless of whether it is
   permitted to initiated new streams.

   Endpoints MUST NOT send data beyond exceed the final
   offset that it has chosen, see Section 10.1.5 for details. limit set by their peer.  An endpoint can receive any type of
   that receives a STREAM frame in this state.  Providing
   flow-control credit using WINDOW_UPDATE frames is necessary to
   continue receiving flow-controlled frames.  In with an ID greater than the limit it has
   sent MUST treat this state, as a receiver
   MAY ignore WINDOW_UPDATE frames for stream error of type
   QUIC_TOO_MANY_OPEN_STREAMS (Section 12), unless this stream, which might arrive
   for is a short period after result of a frame bearing
   change in the FIN flag is sent.

10.1.4.  half-closed (remote) initial offsets (see Section 7.3.2).

   A stream receiver MUST NOT renege on an advertisement; that is "half-closed (remote)" is no longer being used by
   the peer to send any data.  In this state, a sender is no longer
   obligated to maintain is, once a
   receiver stream-level flow-control window.

   A advertises a stream that is in the "half-closed (remote)" state will have ID via a MAX_STREAM_ID frame, it MUST
   NOT subsequently advertise a
   final offset for received data, see Section 10.1.5 for details. smaller maximum ID.  A stream in this state can be used by the endpoint to send sender may
   receive MAX_STREAM_ID frames out of order; a sender MUST therefore
   ignore any type.  In this state, MAX_STREAM_ID that does not increase the endpoint continues to observe
   advertised stream-level maximum.

10.4.  Sending and connection-level flow-control limits
   (Section 11).

   A stream can transition from this state to "closed" by sending a
   frame that contains a FIN flag or when either peer sends Receiving Data

   Once a RST_STREAM
   frame.

10.1.5.  closed

   The "closed" state stream is created, endpoints may use the terminal state.

   An stream to send and
   receive data.  Each endpoint will learn the final offset may send a series of the STREAM frames
   encapsulating data it receives on a stream when it enters until the "half-closed (remote)" or "closed" state.
   The final offset stream is carried explicitly terminated in that
   direction.  Streams are an ordered byte-stream abstraction, and they
   have no other structure within them.  STREAM frame boundaries are not
   expected to be preserved in retransmissions from the RST_STREAM frame;
   otherwise, sender or during
   delivery to the final offset application at the receiver.

   When new data is to be sent on a stream, a sender MUST set the
   encapsulating STREAM frame's offset field to the stream offset of the end
   first byte of this new data.  The first byte of the data
   carried in STREAM frame marked with a FIN flag.

   An endpoint MUST NOT send data that is sent on
   a stream at or beyond has the final
   offset.

   Once a final offset for a stream is known, it cannot change.  If a
   RST_STREAM or STREAM frame causes the final offset to change for a
   stream, an endpoint SHOULD respond with 0.  The largest offset delivered on a
   QUIC_STREAM_DATA_AFTER_TERMINATION error (see Section 12).
   stream MUST be less than 2^64.  A receiver SHOULD treat receipt of MUST ensure that received
   stream data at or beyond is delivered to the final offset application as a QUIC_STREAM_DATA_AFTER_TERMINATION error.  Generating these
   errors an ordered byte-
   stream.  Data received out of order MUST be buffered for later
   delivery, as long as it is not mandatory, but only because requiring that an endpoint
   generate these errors also means that the endpoint needs to maintain in violation of the final offset state for closed streams, which could mean a
   significant state commitment. receiver's flow
   control limits.

   An endpoint that receives a RST_STREAM frame (and which has not sent
   a FIN or a RST_STREAM) MUST immediately respond with a RST_STREAM
   frame, and MUST NOT send any more data on any stream without ensuring that it
   is within the stream.  This endpoint
   may continue receiving frames for data limits set by its peer.  The cryptographic
   handshake stream, Stream 0, is exempt from the stream on which a RST_STREAM connection-level data
   limits established by MAX_DATA.  Stream 0 is
   received.

   If this state still subject to stream-
   level data limits and MAX_STREAM_DATA.

   Flow control is reached as a result of sending a RST_STREAM frame, described in detail in Section 11, and congestion
   control is described in the peer companion document [QUIC-RECOVERY].

10.5.  Stream Prioritization

   Stream multiplexing has a significant effect on application
   performance if resources allocated to streams are correctly
   prioritized.  Experience with other multiplexed protocols, such as
   HTTP/2 [RFC7540], shows that receives the RST_STREAM frame might have already sent -
   or enqueued for sending - effective prioritization strategies have
   a significant positive impact on performance.

   QUIC does not provide frames for exchanging prioritization
   information.  Instead it relies on receiving priority information
   from the stream application that cannot be
   withdrawn.  An endpoint MUST ignore frames uses QUIC.  Protocols that it receives on closed
   streams after it has sent a RST_STREAM frame.  An endpoint MAY choose use QUIC are
   able to limit the period over which it ignores frames and treat frames define any prioritization scheme that arrive after this time suits their application
   semantics.  A protocol might define explicit messages for signaling
   priority, such as being those defined in error.

   STREAM frames received after sending RST_STREAM are counted toward
   the connection and stream flow-control windows.  Even though these
   frames might be ignored, because they are sent before their sender
   receives the RST_STREAM, the sender will consider HTTP/2; it could define rules that
   allow an endpoint to determine priority based on context; or it could
   leave the frames determination to count
   against its flow-control windows.

   In the absence of more specific guidance elsewhere in this document,
   implementations application.

   A QUIC implementation SHOULD treat the receipt of a frame that is not
   expressly permitted provide ways in which an application can
   indicate the description of a state as a connection
   error (Section 12).  Frames relative priority of unknown types are ignored.

   (TODO: QUIC_STREAM_NO_ERROR is a special case.  Write it up.)

10.2.  Stream Identifiers

   Streams are identified streams.  When deciding which
   streams to dedicate resources to, QUIC SHOULD use the information
   provided by an unsigned 32-bit integer, referred the application.  Failure to account for priority of
   streams can result in suboptimal performance.

   Stream priority is most relevant when deciding which stream data will
   be transmitted.  Often, there will be limits on what can be
   transmitted as a result of connection flow control or the StreamID.  To avoid StreamID collision, clients MUST initiate
   streams usinge odd-numbered StreamIDs; streams initiated by current
   congestion controller state.

   Giving preference to the
   server MUST use even-numbered StreamIDs.

   A StreamID transmission of zero (0x0) is reserved its own management frames
   ensures that the protocol functions efficiently.  That is,
   prioritizing frames other than STREAM frames ensures that loss
   recovery, congestion control, and used for connection-level flow control frames (Section 11); the StreamID of zero cannot operate effectively.

   Stream 0 MUST be used prioritized over other streams prior to establish a new stream.

   StreamID 1 (0x1) is reserved for the
   completion of the cryptographic handshake.
   StreamID 1 MUST NOT be used for application data, and MUST be  This includes the
   first client-initiated stream.

   A QUIC endpoint cannot reuse a StreamID on a given connection.
   Streams MUST be created in sequential order.  Open streams can be
   used in
   retransmission of the second flight of client handshake messages,
   that is, the TLS Finished and any order.  Streams client authentication messages.

   STREAM frames that are used out of order result in
   lower-numbered streams determined to be lost SHOULD be retransmitted
   before sending new data, unless application priorities indicate
   otherwise.  Retransmitting lost STREAM frames can fill in gaps, which
   allows the same direction being counted as open.

   All streams, including stream 1, count toward this limit.  Thus, a
   concurrent stream peer to consume already received data and free up flow
   control window.

11.  Flow Control

   It is necessary to limit the amount of 0 will cause data that a sender may have
   outstanding at any time, so as to prevent a fast sender from
   overwhelming a connection slow receiver, or to be unusable.
   Application protocols that use prevent a malicious sender from
   consuming significant resources at a receiver.  This section
   describes QUIC's flow-control mechanisms.

   QUIC might require employs a certain minimum credit-based flow-control scheme similar to HTTP/2's
   flow control [RFC7540].  A receiver advertises the number of streams octets
   it is prepared to function correctly.  If receive on a peer advertises an
   concurrent given stream limit (concurrent_streams) that is too small and for the selected application protocol entire
   connection.  This leads to function, an endpoint MUST
   terminate the connection with an error two levels of type
   QUIC_TOO_MANY_OPEN_STREAMS (Section 12).

10.3. flow control in QUIC: (i)
   Connection flow control, which prevents senders from exceeding a
   receiver's buffer capacity for the connection, and (ii) Stream Concurrency

   An endpoint limits flow
   control, which prevents a single stream from consuming the number of concurrently active incoming streams entire
   receive buffer for a connection.

   A receiver sends MAX_DATA or MAX_STREAM_DATA frames to the sender to
   advertise additional credit by setting sending the concurrent stream limit (see Section 7.3.1) absolute byte offset in
   the
   transport parameters.  The maximum concurrent streams setting connection or stream which it is
   specific to each endpoint and applies only willing to the peer receive.

   A receiver MAY advertise a larger offset at any point by sending
   MAX_DATA or MAX_STREAM_DATA frames.  A receiver MUST NOT renege on an
   advertisement; that receives
   the setting.  That is, clients specify the maximum number of
   concurrent streams the server can initiate, and servers specify the
   maximum number once a receiver advertises an offset, it MUST
   NOT subsequently advertise a smaller offset.  A sender could receive
   MAX_DATA or MAX_STREAM_DATA frames out of concurrent streams the client can initiate.

   Streams order; a sender MUST
   therefore ignore any flow control offset that are in does not move the "open" state or in either of
   window forward.

   A receiver MUST close the "half-
   closed" states count toward connection with a
   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA error (Section 12) if the maximum number of streams that an
   endpoint is permitted to open.  Streams in any of these three states
   count toward
   peer violates the limit advertised in the concurrent connection or stream limit. data limits.

   A recently closed stream sender MUST also be considered to count toward this
   limit until packets containing all send BLOCKED frames required to close the
   stream have been acknowledged.  For a stream which closed cleanly,
   this means all STREAM frames have been acknowledged; for a stream
   which closed abruptly, this means the RST_STREAM frame indicate it has been
   acknowledged.

   Endpoints MUST NOT exceed the limit set data to write
   but is blocked by their peer.  An endpoint
   that receives a STREAM frame that causes its advertised concurrent lack of connection or stream limit flow control credit.
   BLOCKED frames are expected to be exceeded MUST treat this as a stream error of type
   QUIC_TOO_MANY_OPEN_STREAMS (Section 12).

10.4.  Sending sent infrequently in common cases,
   but they are considered useful for debugging and Receiving Data

   Once monitoring purposes.

   A receiver advertises credit for a stream is created, endpoints may by sending a
   MAX_STREAM_DATA frame with the Stream ID set appropriately.  A
   receiver could use the stream current offset of data consumed to determine
   the flow control offset to be advertised.  A receiver MAY send and
   receive data.  Each endpoint may send a series
   MAX_STREAM_DATA frames in multiple packets in order to make sure that
   the sender receives an update before running out of flow control
   credit, even if one of STREAM frames
   encapsulating data on the packets is lost.

   Connection flow control is a stream until limit to the total bytes of stream is terminated data
   sent in that
   direction.  Streams are an ordered byte-stream abstraction, and they
   have no other structure within them. STREAM frame boundaries frames on all streams.  A receiver advertises credit
   for a connection by sending a MAX_DATA frame.  A receiver maintains a
   cumulative sum of bytes received on all streams, which are not
   expected used to be preserved in retransmissions from the sender or during
   delivery
   check for flow control violations.  A receiver might use a sum of
   bytes consumed on all contributing streams to determine the application at the receiver.

   When new maximum
   data is limit to be sent advertised.

11.1.  Edge Cases and Other Considerations

   There are some edge cases which must be considered when dealing with
   stream and connection level flow control.  Given enough time, both
   endpoints must agree on a stream, a sender MUST set flow control state.  If one end believes it
   can send more than the
   encapsulating STREAM frame's offset field other end is willing to receive, the stream offset of the
   first byte of this new data.  The first byte of
   connection will be torn down when too much data that arrives.

   Conversely if a sender believes it is sent on blocked, while endpoint B
   expects more data can be received, then the connection can be in a stream has
   deadlock, with the stream offset 0.  The largest offset delivered on sender waiting for a MAX_DATA or MAX_STREAM_DATA
   frame which will never come.

   On receipt of a RST_STREAM frame, an endpoint will tear down state
   for the matching stream MUST be less than 2^64.  A receiver MUST ensure that received
   stream and ignore further data is delivered to the application as an ordered byte- arriving on that
   stream.  Data received  This could result in the endpoints getting out of sync,
   since the RST_STREAM frame may have arrived out of order MUST and there
   may be buffered for later
   delivery, as long as it is not further bytes in violation of the receiver's flow
   control limits. flight.  The cryptographic handshake stream, Stream 1, MUST NOT be subject to
   congestion control or connection-level flow control, but MUST be
   subject to stream-level flow control.  An endpoint MUST NOT send data
   on any other stream without consulting the congestion controller and sender would have counted
   the data against its connection level flow controller.

   Flow control is described in detail in Section 11, and congestion control is described in the companion document [QUIC-RECOVERY].

10.5.  Stream Prioritization

   Stream multiplexing has budget, but a significant effect on application
   performance if resources allocated to streams are correctly
   prioritized.  Experience with other multiplexed protocols, such as
   HTTP/2 [RFC7540], shows
   receiver that effective prioritization strategies have
   a significant positive impact on performance.

   QUIC does has not provide frames for exchanging priotization information.
   Instead it relies on receiving priority information from the
   application that uses QUIC.  Protocols that use QUIC are able received these bytes would not know to
   define any prioritization scheme that suits their application
   semantics.  A protocol might define explicit messages for signaling
   priority, such include
   them as those defined in HTTP/2; it could define rules that
   allow an endpoint to determine priority based well.  The receiver must learn the number of bytes that were
   sent on context; or it could
   leave the determination stream to make the application.

   A QUIC implementation SHOULD provide ways same adjustment in which an application can
   indicate its connection flow
   controller.

   To avoid this de-synchronization, a RST_STREAM sender MUST include
   the relative priority of streams.  When deciding which
   streams to dedicate resources to, QUIC SHOULD use final byte offset sent on the information
   provided by stream in the application.  Failure RST_STREAM frame.  On
   receiving a RST_STREAM frame, a receiver definitively knows how many
   bytes were sent on that stream before the RST_STREAM frame, and the
   receiver MUST use the final offset to account for priority of
   streams can result in suboptimal performance.

   Stream priority is most relevant when deciding which stream data will
   be transmitted.  Often, there will be limits all bytes sent on what can be
   transmitted as a result of connection flow control or the current
   congestion controller state.

   Giving preference to
   the transmission of stream in its own management frames
   ensures that the protocol functions efficiently.  That is,
   prioritizing frames other than STREAM frames ensures that loss
   recovery, congestion control, and connection level flow control operate effectively.

   Stream 1 MUST be prioritized over other controller.

11.1.1.  Response to a RST_STREAM

   Since streams prior are bidirectional, a sender of a RST_STREAM needs to
   know how many bytes the
   completion of peer has sent on the cryptographic handshake.  This includes stream.  If an endpoint
   receives a RST_STREAM frame and has sent neither a FIN nor a
   RST_STREAM, it MUST send a RST_STREAM in response, bearing the
   retransmission offset
   of the second flight of client handshake messages,
   that is, last byte sent on this stream as the TLS Finished final offset.

11.1.2.  Data Limit Increments

   This document leaves when and any client authentication messages.

   STREAM how many bytes to advertise in a
   MAX_DATA or MAX_STREAM_DATA to implementations, but offers a few
   considerations.  These frames that are determined contribute to be lost SHOULD be retransmitted
   before connection overhead.
   Therefore frequently sending new data, unless application priorities indicate
   otherwise.  Retransmitting lost STREAM frames can fill in gaps, which
   allows with small changes is
   undesirable.  At the peer same time, infrequent updates require larger
   increments to consume already received data and free up flow
   control window.

11.  Flow Control

   It limits if blocking is necessary to limit the amount of data that be avoided.  Thus, larger
   updates require a sender may have
   outstanding at any time, so as receiver to prevent a fast sender from
   overwhelming a slow receiver, or commit to prevent a malicious sender from
   consuming significant resources at larger resource commitments.
   Thus there is a receiver.  This section
   describes QUIC's flow-control mechanisms.

   QUIC employs tradeoff between resource commitment and overhead
   when determining how large a credit-based flow-control scheme similar to HTTP/2's
   flow control [RFC7540]. limit is advertised.

   A receiver advertises MAY use an autotuning mechanism to tune the number of octets frequency and
   amount that it is prepared to receive increases data limits based on a given stream roundtrip time
   estimate and for the entire
   connection.  This leads to two levels of flow control in QUIC: (i)
   Connection flow control, rate at which prevents senders from exceeding a
   receiver's buffer capacity for the connection, and (ii) receiving application consumes
   data, similar to common TCP implementations.

11.2.  Stream Limit Increment

   As with flow control, which prevents a single stream from consuming the entire
   receive buffer for a connection.

   A receiver sends WINDOW_UPDATE frames this document leaves when and how many streams
   to the sender make available to advertise
   additional credit by sending the absolute byte offset in a peer via MAX_STREAM_ID to implementations, but
   offers a few considerations.  MAX_STREAM_ID frames constitute minimal
   overhead, while withholding MAX_STREAM_ID frames can prevent the stream
   or in peer
   from using the connection which it is willing available parallelism.

   Implementations will likely want to receive.

   The initial flow control credit is 65536 bytes for both increase the maximum stream
   and connection flow controllers. ID as
   peer-initiated streams close.  A receiver MAY advertise a larger offset at any point in also advance the
   connection by sending a WINDOW_UPDATE frame.  A receiver MUST NOT
   renege
   maximum stream ID based on an advertisement; that is, once a receiver advertises an
   offset via a WINDOW_UPDATE frame, it MUST NOT subsequently advertise
   a smaller offset.  A sender may receive WINDOW_UPDATE frames out of
   order; current activity, system conditions, and
   other environmental factors.

11.2.1.  Blocking on Flow Control

   If a sender MUST therefore ignore any WINDOW_UPDATE that does not
   move the window forward.

   A receiver MUST close the connection with receive a
   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA error (Section 12) if the
   peer violates the advertised stream MAX_DATA or connection MAX_STREAM_DATA frame when
   it has run out of flow control
   windows.

   A credit, the sender will be blocked and
   MUST send a BLOCKED frames to indicate it has data to write
   but is blocked by lack of connection or stream flow control credit.
   BLOCKED STREAM_BLOCKED frame.  These frames are
   expected to be sent infrequently in common cases,
   but they are considered useful for debugging and monitoring purposes. at the receiver; they do not
   require any other action.  A receiver advertises credit SHOULD NOT wait for a stream by sending a WINDOW_UPDATE BLOCKED
   or STREAM_BLOCKED frame with the StreamID set appropriately.  A receiver may use the
   current offset of data consumed to determine the flow control offset before sending MAX_DATA or MAX_STREAM_DATA,
   since doing so will mean that a sender is unable to be advertised.  A receiver MAY send copies for an
   entire round trip.

   For smooth operation of a WINDOW_UPDATE
   frame in multiple packets in order the congestion controller, it is generally
   considered best to make sure that not let the sender
   receives it before running out of flow control credit, even go into quiescence if one of
   the packets is lost.

   Connection flow control is
   avoidable.  To avoid blocking a limit sender, and to reasonably account for
   the total bytes possibiity of stream data
   sent in STREAM frames on all streams contributing to connection flow
   control.  A receiver advertises credit for loss, a connection by sending receiver should send a
   WINDOW_UPDATE MAX_DATA or
   MAX_STREAM_DATA frame with at least two roundtrips before it expects the StreamID set
   sender to zero (0x00). get blocked.

   A receiver
   maintains sender sends a cumulative sum of bytes received on all streams
   contributing to connection-level flow control, to check for flow
   control violations.  A receiver may maintain single BLOCKED or STREAM_BLOCKED frame only once
   when it reaches a cumulative sum of
   bytes consumed on all contributing streams to determine data limit.  A sender MUST NOT send multiple
   BLOCKED or STREAM_BLOCKED frames for the
   connection-level flow control offset same data limit, unless the
   original frame is determined to be advertised.

11.1.  Edge Cases and Other Considerations

   There are some edge cases which must be considered when dealing with
   stream and connection level flow control.  Given enough time, both
   endpoints must agree on flow control state.  If one end believes it lost.  Another BLOCKED or
   STREAM_BLOCKED frame can send more than be sent after the other end data limit is willing to receive, increased.

11.3.  Stream Final Offset

   The final offset is the count of the
   connection will be torn down when too much data arrives.  Conversely
   if number of octets that are
   transmitted on a sender believes it stream.  For a stream that is blocked, while endpoint B expects more
   data can be received, then reset, the connection can be final
   offset is carried explicitly in a deadlock, with the sender waiting for a WINDOW_UPDATE which will never come.

11.1.1.  Mid-stream RST_STREAM

   On receipt frame.  Otherwise, the
   final offset is the offset of the end of the data carried in STREAM
   frame marked with a RST_STREAM frame, an FIN flag.

   An endpoint will tear down state know the final offset for a stream when the matching stream and ignore further
   enters the "half-closed (remote)" state.  However, if there is
   reordering or loss, an endpoint might learn the final offset prior to
   entering this state if it is carried on a STREAM frame.

   An endpoint MUST NOT send data arriving on that
   stream.  This could result in the endpoints getting out of sync,
   since a stream at or beyond the final
   offset.

   Once a final offset for a stream is known, it cannot change.  If a
   RST_STREAM or STREAM frame may have arrived out causes the final offset to change for a
   stream, an endpoint SHOULD respond with a
   QUIC_STREAM_DATA_AFTER_TERMINATION error (see Section 12).  A
   receiver SHOULD treat receipt of order and there
   may be further bytes in flight.  The data sender would have counted at or beyond the data against its connection level flow control budget, but final offset
   as a
   receiver that has QUIC_STREAM_DATA_AFTER_TERMINATION error, even after a stream is
   closed.  Generating these errors is not received mandatory, but only because
   requiring that an endpoint generate these bytes would not know errors also means that the
   endpoint needs to include
   them as well.  The receiver must learn maintain the number of bytes final offset state for closed streams,
   which could mean a significant state commitment.

12.  Error Handling

   An endpoint that were
   sent on detects an error SHOULD signal the stream existence of that
   error to make the same adjustment in its peer.  Errors can affect an entire connection flow
   controller.

   To avoid this de-synchronization, (see
   Section 12.1), or a RST_STREAM sender MUST include
   the final byte offset sent on the single stream (see Section 12.2).

   The most appropriate error code (Section 12.3) SHOULD be included in
   the RST_STREAM frame.  On
   receiving a RST_STREAM frame, a receiver definitively knows how many
   bytes were sent on frame that stream before signals the RST_STREAM frame, and error.  Where this specification
   identifies error conditions, it also identifies the
   receiver error code that
   is used.

   Public Reset is not suitable for any error that can be signaled with
   a CONNECTION_CLOSE or RST_STREAM frame.  Public Reset MUST use NOT be
   sent by an endpoint that has the final offset state necessary to account for all bytes sent send a frame on
   the stream connection.

12.1.  Connection Errors

   Errors that result in its the connection level flow controller.

11.1.2.  Response to a RST_STREAM

   Since streams are bidirectional, a sender being unusable, such as an
   obvious violation of a RST_STREAM needs to
   know how many bytes the peer has sent on the stream.  If protocol semantics or corruption of state that
   affects an entire connection, MUST be signaled using a
   CONNECTION_CLOSE frame (Section 8.13).  An endpoint
   receives MAY close the
   connection in this manner, even if the error only affects a RST_STREAM single
   stream.

   A CONNECTION_CLOSE frame and has could be sent neither in a FIN nor packet that is lost.  An
   endpoint SHOULD be prepared to retransmit a
   RST_STREAM, packet containing a
   CONNECTION_CLOSE frame if it MUST send receives more packets on a RST_STREAM in response, bearing terminated
   connection.  Limiting the offset number of retransmissions and the last byte sent on time over
   which this stream as the final offset.

11.1.3.  Offset Increment

   This document leaves when and how many bytes to advertise in a
   WINDOW_UPDATE to the implementation, but offers a few considerations.
   WINDOW_UPDATE frames constitute overhead, and therefore, sending a
   WINDOW_UPDATE with small offset increments packet is undesirable.  At the
   same time, sending WINDOW_UPDATES with large offset increments
   requires sent limits the sender to commit to effort expended on
   terminated connections.

   An endpoint that amount of buffer.

   Implementations must find the correct tradeoff between these sides to
   determine how large an offset increment chooses not to send in retransmit packets containing
   CONNECTION_CLOSE risks a WINDOW_UPDATE.

   A receiver MAY use an autotuning mechanism to tune the size of peer missing the
   offset increment first such packet.  The
   only mechanism available to advertise based on a roundtrip time estimate and
   the rate at which the receiving application consumes data, similar an endpoint that continues to
   common TCP implementations.

11.1.4.  BLOCKED frames

   If a sender does not receive
   data for a WINDOW_UPDATE frame when it has run
   out of flow control credit, the sender will be blocked and MUST send
   a BLOCKED frame.  A BLOCKED frame terminated connection is expected to be useful for
   debugging at the receiver.  A receiver SHOULD NOT wait for send a Public Reset packet.

12.2.  Stream Errors

   If the error affects a BLOCKED
   frame before sending single stream, but otherwise leaves the
   connection in a WINDOW_UPDATE, since doing so will cause at
   least one roundtrip of quiescence.  For smooth operation of recoverable state, the
   congestion controller, it is generally considered best endpoint can sent a RST_STREAM
   frame (Section 8.9) with an appropriate error code to not let terminate just
   the
   sender go into quiescence if avoidable.  To avoid blocking a sender,
   and affected stream.

   Stream 0 is critical to reasonably account for the possibiity functioning of loss, a receiver
   should send the entire connection.  If
   stream 0 is closed with either a WINDOW_UPDATE RST_STREAM or STREAM frame at least two roundtrips before it
   expects bearing
   the sender to get blocked.

12.  Error Handling

   An endpoint that detects FIN flag, an endpoint MUST generate a connection error SHOULD signal the existence of type
   QUIC_CLOSED_CRITICAL_STREAM.

   Some application protocols make other streams critical to that
   error
   protocol.  An application protocol does not need to its peer.  Errors can affect an entire connection (see
   Section 12.1), or inform the
   transport that a single stream (see Section 12.2).

   The most is critical; it can instead generate
   appropriate error code (Section 12.3) SHOULD be included errors in response to being notified that the critical
   stream is closed.

   An endpoint MAY send a RST_STREAM frame that signals in the error.  Where this specification
   identifies error conditions, it also identifies same packet as a
   CONNECTION_CLOSE frame.

12.3.  Error Codes

   Error codes are 32 bits long, with the first two bits indicating the
   source of the error code that
   is used.

   Public Reset is not suitable code:

   0x00000000-0x3FFFFFFF:  Application-specific error codes.  Defined by
      each application-layer protocol.

   0x40000000-0x7FFFFFFF:  Reserved for any host-local error that can be signaled with
   a CONNECTION_CLOSE or RST_STREAM frame.  Public Reset codes.  These
      codes MUST NOT be sent by an endpoint that has the state necessary to send a frame on
   the connection.

12.1.  Connection Errors

   Errors that result peer, but MAY be used in the connection being unusable, such as an
   obvious violation of protocol semantics or corruption API return
      codes and logs.

   0x80000000-0xBFFFFFFF:  QUIC transport error codes, including packet
      protection errors.  Applicable to all uses of state that
   affects an entire connection, MUST be signaled using a
   CONNECTION_CLOSE frame (Section 8.8).  An endpoint MAY close QUIC.

   0xC0000000-0xFFFFFFFF:  Cryptographic error codes.  Defined by the
   connection
      cryptographic handshake protocol in this manner, even if use.

   This section lists the defined QUIC transport error only affects a single
   stream.

   A CONNECTION_CLOSE frame could be sent in a packet codes that is lost.  An
   endpoint SHOULD may be prepared to retransmit a packet containing
   used in a CONNECTION_CLOSE frame if it receives more packets on or RST_STREAM frame.  Error codes share a terminated
   connection.  Limiting
   common code space.  Some error codes apply only to either streams or
   the number of retransmissions entire connection and have no defined semantics in the time over
   which this final packet is sent limits other
   context.

   QUIC_INTERNAL_ERROR (0x80000001):  Connection has reached an invalid
      state.

   QUIC_STREAM_DATA_AFTER_TERMINATION (0x80000002):  There were data
      frames after the effort expended on
   terminated connections.

   An endpoint that chooses not to retransmit packets containing
   CONNECTION_CLOSE risks a peer missing fin or reset.

   QUIC_INVALID_PACKET_HEADER (0x80000003):  Control frame is malformed.

   QUIC_INVALID_FRAME_DATA (0x80000004):  Frame data is malformed.

   QUIC_MULTIPLE_TERMINATION_OFFSETS (0x80000005):  Multiple final
      offset values were received on the first such packet.  The
   only mechanism available to an endpoint same stream

   QUIC_STREAM_CANCELLED (0x80000006):  The stream was cancelled

   QUIC_CLOSED_CRITICAL_STREAM (0x80000007):  A stream that continues to receive
   data for a terminated connection is critical
      to send a Public Reset packet.

12.2.  Stream Errors

   If the error affects a single stream, but otherwise leaves the
   connection in protocol was closed.

   QUIC_MISSING_PAYLOAD (0x80000030):  The packet contained no payload.

   QUIC_INVALID_STREAM_DATA (0x8000002E):  STREAM frame data is
      malformed.

   QUIC_UNENCRYPTED_STREAM_DATA (0x8000003D):  Received STREAM frame
      data is not encrypted.

   QUIC_MAYBE_CORRUPTED_MEMORY (0x80000059):  Received a recoverable state, frame which is
      likely the endpoint can sent a result of memory corruption.

   QUIC_INVALID_RST_STREAM_DATA (0x80000006):  RST_STREAM frame (Section 8.5) with data is
      malformed.

   QUIC_INVALID_CONNECTION_CLOSE_DATA (0x80000007):  CONNECTION_CLOSE
      frame data is malformed.

   QUIC_INVALID_GOAWAY_DATA (0x80000008):  GOAWAY frame data is
      malformed.

   QUIC_INVALID_WINDOW_UPDATE_DATA (0x80000039):  WINDOW_UPDATE frame
      data is malformed.

   QUIC_INVALID_BLOCKED_DATA (0x8000003A):  BLOCKED frame data is
      malformed.

   QUIC_INVALID_PATH_CLOSE_DATA (0x8000004E):  PATH_CLOSE frame data is
      malformed.

   QUIC_INVALID_ACK_DATA (0x80000009):  ACK frame data is malformed.

   QUIC_INVALID_VERSION_NEGOTIATION_PACKET (0x8000000A):  Version
      negotiation packet is malformed.

   QUIC_INVALID_PUBLIC_RST_PACKET (0x8000000b):  Public RST packet is
      malformed.

   QUIC_DECRYPTION_FAILURE (0x8000000c):  There was an appropriate error code to terminate just
   the affected stream.

   Stream 1 is critical to the functioning of the entire connection.  If
   stream 1 decrypting.

   QUIC_ENCRYPTION_FAILURE (0x8000000d):  There was an error encrypting.

   QUIC_PACKET_TOO_LARGE (0x8000000e):  The packet exceeded
      kMaxPacketSize.

   QUIC_PEER_GOING_AWAY (0x80000010):  The peer is closed with either going away.  May be a RST_STREAM
      client or STREAM frame bearing server.

   QUIC_INVALID_STREAM_ID (0x80000011):  A stream ID was invalid.

   QUIC_INVALID_PRIORITY (0x80000031):  A priority was invalid.

   QUIC_TOO_MANY_OPEN_STREAMS (0x80000012):  Too many streams already
      open.

   QUIC_TOO_MANY_AVAILABLE_STREAMS (0x8000004c):  The peer created too
      many available streams.

   QUIC_PUBLIC_RESET (0x80000013):  Received public reset for this
      connection.

   QUIC_INVALID_VERSION (0x80000014):  Invalid protocol version.

   QUIC_INVALID_HEADER_ID (0x80000016):  The Header ID for a stream was
      too far from the FIN flag, previous.

   QUIC_INVALID_NEGOTIATED_VALUE (0x80000017):  Negotiable parameter
      received during handshake had invalid value.

   QUIC_DECOMPRESSION_FAILURE (0x80000018):  There was an endpoint MUST generate a connection error of type
   QUIC_CLOSED_CRITICAL_STREAM.

   Some application protocols make other streams critical to that
   protocol.  An application protocol does not need
      decompressing data.

   QUIC_NETWORK_IDLE_TIMEOUT (0x80000019):  The connection timed out due
      to inform no network activity.

   QUIC_HANDSHAKE_TIMEOUT (0x80000043):  The connection timed out
      waiting for the
   transport that a stream is critical; it can instead generate
   appropriate errors in response handshake to being notified that the critical
   stream is closed.

   An endpoint MAY send a RST_STREAM frame in the same packet as a
   CONNECTION_CLOSE frame.

12.3.  Error Codes

   Error codes are 32 bits long, with the first two bits indicating the
   source of the error code:

   0x00000000-0x3FFFFFFF:  Application-specific complete.

   QUIC_ERROR_MIGRATING_ADDRESS (0x8000001a):  There was an error codes.  Defined by
      each application-layer protocol.

   0x40000000-0x7FFFFFFF:  Reserved for host-local
      encountered migrating addresses.

   QUIC_ERROR_MIGRATING_PORT (0x80000056):  There was an error codes.  These
      codes MUST NOT be sent to
      encountered migrating port only.

   QUIC_EMPTY_STREAM_FRAME_NO_FIN (0x80000032):  We received a peer, but MAY be used in API return
      codes
      STREAM_FRAME with no data and logs.

   0x80000000-0xBFFFFFFF: no fin flag set.

   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA (0x8000003b):  The peer
      received too much data, violating flow control.

   QUIC_FLOW_CONTROL_SENT_TOO_MUCH_DATA (0x8000003f):  The peer sent too
      much data, violating flow control.

   QUIC_FLOW_CONTROL_INVALID_WINDOW (0x80000040):  The peer received an
      invalid flow control window.

   QUIC_CONNECTION_IP_POOLED (0x8000003e):  The connection has been IP
      pooled into an existing connection.

   QUIC_TOO_MANY_OUTSTANDING_SENT_PACKETS (0x80000044):  The connection
      has too many outstanding sent packets.

   QUIC_TOO_MANY_OUTSTANDING_RECEIVED_PACKETS (0x80000045):  The
      connection has too many outstanding received packets.

   QUIC_CONNECTION_CANCELLED (0x80000046):  The QUIC connection has been
      cancelled.

   QUIC_BAD_PACKET_LOSS_RATE (0x80000047):  Disabled QUIC transport error codes, including because of
      high packet
      protection errors.  Applicable to all uses loss rate.

   QUIC_PUBLIC_RESETS_POST_HANDSHAKE (0x80000049):  Disabled QUIC
      because of QUIC.

   0xC0000000-0xFFFFFFFF:  Cryptographic error codes.  Defined by the
      cryptographic handshake protocol in use.

   This section lists the defined too many PUBLIC_RESETs post handshake.

   QUIC_TIMEOUTS_WITH_OPEN_STREAMS (0x8000004a):  Disabled QUIC transport error codes that may be
   used in a CONNECTION_CLOSE or RST_STREAM frame.  Error codes share a
   common code space.  Some error codes apply only to either because
      of too many timeouts with streams or open.

   QUIC_TOO_MANY_RTOS (0x80000055):  QUIC timed out after too many RTOs.

   QUIC_ENCRYPTION_LEVEL_INCORRECT (0x8000002c):  A packet was received
      with the entire wrong encryption level (i.e. it should have been
      encrypted but was not.)

   QUIC_VERSION_NEGOTIATION_MISMATCH (0x80000037):  This connection and
      involved a version negotiation which appears to have no defined semantics in the other
   context.

   QUIC_INTERNAL_ERROR (0x80000001): been tampered
      with.

   QUIC_IP_ADDRESS_CHANGED (0x80000050):  IP address changed causing
      connection close.

   QUIC_ADDRESS_VALIDATION_FAILURE (0x80000051):  Client address
      validation failed.

   QUIC_TOO_MANY_FRAME_GAPS (0x8000005d):  Stream frames arrived too
      discontiguously so that stream sequencer buffer maintains too many
      gaps.

   QUIC_TOO_MANY_SESSIONS_ON_SERVER (0x80000060):  Connection has reached closed
      because server hit max number of sessions allowed.

13.  Security and Privacy Considerations

13.1.  Spoofed ACK Attack

   An attacker receives an invalid
      state.

   QUIC_STREAM_DATA_AFTER_TERMINATION (0x80000002):  There were data
      frames after STK from the a fin or reset.

   QUIC_INVALID_PACKET_HEADER (0x80000003):  Control frame is malformed.

   QUIC_INVALID_FRAME_DATA (0x80000004):  Frame data is malformed.

   QUIC_MULTIPLE_TERMINATION_OFFSETS (0x80000005):  Multiple final
      offset values were server and then releases the IP
   address on which it received the STK.  The attacker may, in the
   future, spoof this same address (which now presumably addresses a
   different endpoint), and initiate a 0-RTT connection with a server on
   the same stream

   QUIC_STREAM_CANCELLED (0x80000006): victim's behalf.  The stream was cancelled

   QUIC_CLOSED_CRITICAL_STREAM (0x80000007):  A stream that is critical attacker then spoofs ACK frames to the protocol was closed.

   QUIC_MISSING_PAYLOAD (0x80000030):  The packet contained no payload.

   QUIC_INVALID_STREAM_DATA (0x8000002E):  STREAM frame data is
      malformed.

   QUIC_UNENCRYPTED_STREAM_DATA (0x8000003D):  Received STREAM frame
      data is not encrypted.

   QUIC_MAYBE_CORRUPTED_MEMORY (0x80000059):  Received a frame
   server which is
      likely cause the result of memory corruption.

   QUIC_INVALID_RST_STREAM_DATA (0x80000006):  RST_STREAM frame data is
      malformed.

   QUIC_INVALID_CONNECTION_CLOSE_DATA (0x80000007):  CONNECTION_CLOSE
      frame data is malformed.

   QUIC_INVALID_GOAWAY_DATA (0x80000008):  GOAWAY frame data is
      malformed.

   QUIC_INVALID_WINDOW_UPDATE_DATA (0x80000039):  WINDOW_UPDATE frame
      data is malformed.

   QUIC_INVALID_BLOCKED_DATA (0x8000003A):  BLOCKED frame data is
      malformed.

   QUIC_INVALID_PATH_CLOSE_DATA (0x8000004E):  PATH_CLOSE frame data is
      malformed.

   QUIC_INVALID_ACK_DATA (0x80000009):  ACK frame data is malformed.

   QUIC_INVALID_VERSION_NEGOTIATION_PACKET (0x8000000A):  Version
      negotiation packet is malformed.

   QUIC_INVALID_PUBLIC_RST_PACKET (0x8000000b):  Public RST packet is
      malformed.

   QUIC_DECRYPTION_FAILURE (0x8000000c):  There was an error decrypting.

   QUIC_ENCRYPTION_FAILURE (0x8000000d): server to potentially drown the victim in
   data.

   There was an error encrypting.

   QUIC_PACKET_TOO_LARGE (0x8000000e):  The packet exceeded
      kMaxPacketSize.

   QUIC_PEER_GOING_AWAY (0x80000010): are two possible mitigations to this attack.  The peer simplest one
   is going away.  May be that a server can unilaterally create a gap in packet-number
   space.  In the non-attack scenario, the client or server.

   QUIC_INVALID_STREAM_ID (0x80000011):  A stream ID was invalid.

   QUIC_INVALID_PRIORITY (0x80000031):  A priority was invalid.

   QUIC_TOO_MANY_OPEN_STREAMS (0x80000012):  Too many streams already
      open.

   QUIC_TOO_MANY_AVAILABLE_STREAMS (0x8000004c):  The peer created too
      many available streams.

   QUIC_PUBLIC_RESET (0x80000013):  Received public reset will send an ACK frame
   with the larger value for this
      connection.

   QUIC_INVALID_VERSION (0x80000014):  Invalid protocol version.

   QUIC_INVALID_HEADER_ID (0x80000016):  The Header ID largest acknowledged.  In the attack
   scenario, the attacker could acknowledge a packet in the gap.  If the
   server sees an acknowledgment for a stream packet that was
      too far from never sent, the previous.

   QUIC_INVALID_NEGOTIATED_VALUE (0x80000017):  Negotiable parameter
      received during handshake had invalid value.

   QUIC_DECOMPRESSION_FAILURE (0x80000018):  There was an error
      decompressing data.

   QUIC_NETWORK_IDLE_TIMEOUT (0x80000019):  The
   connection timed out due
      to no network activity.

   QUIC_HANDSHAKE_TIMEOUT (0x80000043): can be aborted.

   The connection timed out
      waiting second mitigation is that the server can require that
   acknowledgments for sent packets match the encryption level of the handshake to complete.

   QUIC_ERROR_MIGRATING_ADDRESS (0x8000001a):  There was an error
      encountered migrating addresses.

   QUIC_ERROR_MIGRATING_PORT (0x80000056):  There was an error
      encountered migrating port only.

   QUIC_EMPTY_STREAM_FRAME_NO_FIN (0x80000032):  We received a
      STREAM_FRAME with no data and no fin flag set.

   QUIC_FLOW_CONTROL_RECEIVED_TOO_MUCH_DATA (0x8000003b):  The peer
      received too much data, violating flow control.

   QUIC_FLOW_CONTROL_SENT_TOO_MUCH_DATA (0x8000003f):  The peer
   sent too
      much data, violating flow control.

   QUIC_FLOW_CONTROL_INVALID_WINDOW (0x80000040):  The peer received an
      invalid flow control window.

   QUIC_CONNECTION_IP_POOLED (0x8000003e):  The packet.  This mitigation is useful if the connection has been IP
      pooled into an existing
   ephemeral forward-secure key that is generated and used for every new
   connection.

   QUIC_TOO_MANY_OUTSTANDING_SENT_PACKETS (0x80000044):  The connection
      has too many outstanding  If a packet sent packets.

   QUIC_TOO_MANY_OUTSTANDING_RECEIVED_PACKETS (0x80000045):  The
      connection has too many outstanding is protected with a forward-secure key,
   then any acknowledgments that are received packets.

   QUIC_CONNECTION_CANCELLED (0x80000046): for them MUST also be
   forward-secure protected.  Since the attacker will not have the
   forward secure key, the attacker will not be able to generate
   forward-secure protected packets with ACK frames.

13.2.  Slowloris Attacks

   The attacks commonly known as Slowloris [SLOWLORIS] try to keep many
   connections to the target endpoint open and hold them open as long as
   possible.  These attacks can be executed against a QUIC connection has been
      cancelled.

   QUIC_BAD_PACKET_LOSS_RATE (0x80000047):  Disabled QUIC because endpoint by
   generating the minimum amount of activity necessary to avoid being
   closed for inactivity.  This might involve sending small amounts of
   data, gradually opening flow control windows in order to control the
   sender rate, or manufacturing ACK frames that simulate a high packet loss
   rate.

   QUIC_PUBLIC_RESETS_POST_HANDSHAKE (0x80000049):  Disabled QUIC
      because of too many PUBLIC_RESETs post handshake.

   QUIC_TIMEOUTS_WITH_OPEN_STREAMS (0x8000004a):  Disabled

   QUIC because deployments SHOULD provide mitigations for the Slowloris
   attacks, such as increasing the maximum number of too many timeouts with streams open.

   QUIC_TOO_MANY_RTOS (0x80000055):  QUIC timed out after too many RTOs.

   QUIC_ENCRYPTION_LEVEL_INCORRECT (0x8000002c):  A packet was received
      with clients the wrong encryption level (i.e. it should have been
      encrypted but was not.)

   QUIC_VERSION_NEGOTIATION_MISMATCH (0x80000037):  This connection
      involved server
   will allow, limiting the number of connections a version negotiation which appears to have been tampered
      with.

   QUIC_IP_ADDRESS_CHANGED (0x80000050): single IP address changed causing is
   allowed to make, imposing restrictions on the minimum transfer speed
   a connection close.

   QUIC_ADDRESS_VALIDATION_FAILURE (0x80000051):  Client address
      validation failed.

   QUIC_TOO_MANY_FRAME_GAPS (0x8000005d):  Stream frames arrived too
      discontiguously so that stream sequencer buffer maintains too many
      gaps.

   QUIC_TOO_MANY_SESSIONS_ON_SERVER (0x80000060):  Connection closed
      because server hit max number of sessions allowed.

13.  Security is allowed to have, and Privacy Considerations

13.1.  Spoofed ACK Attack

   An attacker receives an STK from restricting the server length of time
   an endpoint is allowed to stay connected.

13.3.  Stream Fragmentation and then releases Reassembly Attacks

   An adversarial endpoint might intentionally fragment the IP
   address data on which it received the STK.  The attacker may,
   stream buffers in the
   future, spoof this same address (which now presumably addresses order to cause disproportionate memory commitment.
   An adversarial endpoint could open a
   different endpoint), stream and send some STREAM
   frames containing arbitrary fragments of the stream content.

   The attack is mitigated if flow control windows correspond to
   available memory.  However, some receivers will over-commit memory
   and initiate a 0-RTT connection with a server on advertise flow control offsets in the victim's behalf. aggregate that exceed
   actual available memory.  The attacker then spoofs ACK frames to the
   server which cause the server over-commitment strategy can lead to potentially drown the victim in
   data.

   There
   better performance when endpoints are two possible mitigations well behaved, but renders
   endpoints vulnerable to this the stream fragmentation attack.  The simplest one
   is that a server can unilaterally create a gap in packet-number
   space.  In

   QUIC deployments SHOULD provide mitigations against the non-attack scenario, stream
   fragmentation attack.  Mitigations could consist of avoiding over-
   committing memory, delaying reassembly of STREAM frames, implementing
   heuristics based on the client will send age and duration of reassembly holes, or some
   combination.

13.4.  Stream Commitment Attack

   An adversarial endpoint can open lots of streams, exhausting state on
   an ACK frame
   with the larger value for largest acknowledged.  In the attack
   scenario, the attacker endpoint.  The adversarial endpoint could acknowledge repeat the process on a packet
   large number of connections, in the gap.  If the
   server sees an acknowledgment for a packet that was never sent, the
   connection can manner similar to SYN flooding
   attacks in TCP.

   Normally, clients will open streams sequentially, as explained in
   Section 10.1.  However, when several streams are initiated at short
   intervals, transmission error may cause STREAM DATA frames opening
   streams to be aborted.

   The second mitigation is that the server can require that
   acknowledgments for sent packets match the encryption level received out of the
   sent packet.  This mitigation sequence.  A receiver is useful obligated to
   open intervening streams if the connection has an
   ephemeral forward-secure key that a higher-numbered stream ID is generated and used for every received.
   Thus, on a new
   connection. connection, opening stream 2000001 opens 1 million
   streams, as required by the specification.

   The number of active streams is limited by the concurrent stream
   limit transport parameter, as explained in Section 10.3.  If a packet sent is encrypted with a forward-secure key,
   then any acknowledgments that are received for them MUST also be
   forward-secure encrypted.  Since chosen
   judisciously, this limit mitigates the attacker will not have effect of the
   forward secure key, stream
   commitment attack.  However, setting the attacker will not be able limit too low could affect
   performance when applications expect to generate
   forward-secure encrypted packets with ACK frames. open large number of streams.

14.  IANA Considerations

14.1.  QUIC Transport Parameter Registry

   IANA [SHALL add/has added] a registry for "QUIC Transport Parameters"
   under a "QUIC Protocol" heading.

   The "QUIC Transport Parameters" registry governs a 16-bit space.
   This space is split into two spaces that are governed by different
   policies.  Values with the first byte in the range 0x00 to 0xfe (in
   hexadecimal) are assigned via the Specification Required policy
   [RFC5226].  Values with the first byte 0xff are reserved for Private
   Use [RFC5226].

   Registrations MUST include the following fields:

   Value:  The numeric value of the assignment (registrations will be
      between 0x0000 and 0xfeff).

   Parameter Name:  A short mnemonic for the parameter.

   Specification:  A reference to a publicly available specification for
      the value.

   The nominated expert(s) verify that a specification exists and is
   readily accessible.  The expert(s) are encouraged to be biased
   towards approving registrations unless they are abusive, frivolous,
   or actively harmful (not merely aesthetically displeasing, or
   architecturally dubious).

   The initial contents of this registry are shown in Table 4.

            +--------+------------------------+---------------+

           +--------+-------------------------+---------------+
           | Value  | Parameter Name          | Specification |
            +--------+------------------------+---------------+
           +--------+-------------------------+---------------+
           | 0x0000 | stream_fc_offset initial_max_stream_data | Section 7.3.1 |
           |        |                         |               |
           | 0x0001 | connection_fc_offset initial_max_data        | Section 7.3.1 |
           |        |                         |               |
           | 0x0002 | concurrent_streams initial_max_stream_id   | Section 7.3.1 |
           |        |                         |               |
           | 0x0003 | idle_timeout            | Section 7.3.1 |
           |        |                         |               |
           | 0x0004 | truncate_connection_id  | Section 7.3.1 |
            +--------+------------------------+---------------+
           +--------+-------------------------+---------------+

            Table 4: Initial QUIC Transport Parameters Entries

15.  References

15.1.  Normative References

   [I-D.ietf-tls-tls13]
              Rescorla, E., "The Transport Layer Security (TLS) Protocol
              Version 1.3", draft-ietf-tls-tls13-19 draft-ietf-tls-tls13-20 (work in progress),
              March
              April 2017.

   [QUIC-RECOVERY]
              Iyengar, J., Ed. and I. Swett, Ed., "QUIC Loss Detection
              and Congestion Control". Control", draft-ietf-quic-recovery (work in
              progress), May 2017.

   [QUIC-TLS]
              Thomson, M., Ed. and S. Turner, Ed., "Using Transport
              Layer Security (TLS) to Secure QUIC". QUIC", draft-ietf-quic-tls
              (work in progress), May 2017.

   [RFC1191]  Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191,
              DOI 10.17487/RFC1191, November 1990,
              <http://www.rfc-editor.org/info/rfc1191>.

   [RFC1981]  McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery
              for IP version 6", RFC 1981, DOI 10.17487/RFC1981, August
              1996, <http://www.rfc-editor.org/info/rfc1981>.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <http://www.rfc-editor.org/info/rfc2119>.

   [RFC3629]  Yergeau, F., "UTF-8, a transformation format of ISO
              10646", STD 63, RFC 3629, DOI 10.17487/RFC3629, November
              2003, <http://www.rfc-editor.org/info/rfc3629>.

   [RFC4821]  Mathis, M. and J. Heffner, "Packetization Layer Path MTU
              Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007,
              <http://www.rfc-editor.org/info/rfc4821>.

   [RFC5226]  Narten, T. and H. Alvestrand, "Guidelines for Writing an
              IANA Considerations Section in RFCs", BCP 26, RFC 5226,
              DOI 10.17487/RFC5226, May 2008,
              <http://www.rfc-editor.org/info/rfc5226>.

15.2.  Informative References

   [EARLY-DESIGN]
              Roskind, J., "QUIC: Multiplexed Transport Over UDP",
              December 2013, <https://goo.gl/dMVtFi>.

   [RFC2360]  Scott, G., "Guide for Internet Standards Writers", BCP 22,
              RFC 2360, DOI 10.17487/RFC2360, June 1998,
              <http://www.rfc-editor.org/info/rfc2360>.

   [RFC4086]  Eastlake 3rd, D., Schiller, J., and S. Crocker,
              "Randomness Requirements for Security", BCP 106, RFC 4086,
              DOI 10.17487/RFC4086, June 2005,
              <http://www.rfc-editor.org/info/rfc4086>.

   [RFC6824]  Ford, A., Raiciu, C., Handley, M., and O. Bonaventure,
              "TCP Extensions for Multipath Operation with Multiple
              Addresses", RFC 6824, DOI 10.17487/RFC6824, January 2013,
              <http://www.rfc-editor.org/info/rfc6824>.

   [RFC7301]  Friedl, S., Popov, A., Langley, A., and E. Stephan,
              "Transport Layer Security (TLS) Application-Layer Protocol
              Negotiation Extension", RFC 7301, DOI 10.17487/RFC7301,
              July 2014, <http://www.rfc-editor.org/info/rfc7301>.

   [RFC7540]  Belshe, M., Peon, R., and M. Thomson, Ed., "Hypertext
              Transfer Protocol Version 2 (HTTP/2)", RFC 7540,
              DOI 10.17487/RFC7540, May 2015,
              <http://www.rfc-editor.org/info/rfc7540>.

   [SLOWLORIS]
              RSnake Hansen, R., "Welcome to Slowloris...", June 2009,
              <https://web.archive.org/web/20150315054838/
              http://ha.ckers.org/slowloris/>.

   [SST]      Ford, B., "Structured Streams: A New Transport
              Abstraction", DOI 10.1145/1282427.1282421, streams", ACM SIGCOMM Computer
              Communication Review Volume 37 Issue 4, Vol. 37, pp. 361,
              DOI 10.1145/1282427.1282421, October 2007.

15.3.  URIs

   [1] https://github.com/quicwg/base-drafts/wiki/QUIC-Versions

Appendix A.  Contributors

   The original authors of this specification were Ryan Hamilton, Jana
   Iyengar, Ian Swett, and Alyssa Wilk.

   The original design and rationale behind this protocol draw
   significantly from work by Jim Roskind [EARLY-DESIGN].  In
   alphabetical order, the contributors to the pre-IETF QUIC project at
   Google are: Britt Cyr, Jeremy Dorfman, Ryan Hamilton, Jana Iyengar,
   Fedor Kouranov, Charles Krasic, Jo Kulik, Adam Langley, Jim Roskind,
   Robbie Shade, Satyam Shekhar, Cherie Shi, Ian Swett, Raman Tenneti,
   Victor Vasiliev, Antonio Vicente, Patrik Westin, Alyssa Wilk, Dale
   Worley, Fan Yang, Dan Zhang, Daniel Ziegler.

Appendix B.  Acknowledgments

   Special thanks are due to the following for helping shape pre-IETF
   QUIC and its deployment: Chris Bentzel, Misha Efimov, Roberto Peon,
   Alistair Riddoch, Siddharth Vijayakrishnan, and Assar Westerlund.

   This document has benefited immensely from various private
   discussions and public ones on the quic@ietf.org and proto-
   quic@chromium.org mailing lists.  Our thanks to all.

Appendix C.  Change Log

      *RFC Editor's Note:* Please remove this section prior to
      publication of a final version of this document.

   Issue and pull request numbers are listed with a leading octothorp.

C.1.  Since draft-ietf-quic-transport-01: draft-ietf-quic-transport-02

   o  The size of the initial packet payload has a fixed minimum (#267,
      #472)

   o  Define when Version Negotiation packets are ignored (#284, #294,
      #241, #143, #474)

   o  The 64-bit FNV-1a algorithm is used for integrity protection of
      unprotected packets (#167, #480, #481, #517)

   o  Rework initial packet types to change how the connection ID is
      chosen (#482, #442, #493)

   o  No timestamps are forbidden in unprotected packets (#542, #429)

   o  Cryptographic handshake is now on stream 0 (#456)

   o  Remove congestion control exemption for cryptographic handshake
      (#248, #476)

   o  Version 1 of QUIC uses TLS; a new version is needed to use a
      different handshake protocol (#516)

   o  STREAM frames have a reduced number of offset lengths (#543, #430)

   o  Split some frames into separate connection- and stream- level
      frames (#443)

      *  WINDOW_UPDATE split into MAX_DATA and MAX_STREAM_DATA (#450)

      *  BLOCKED split to match WINDOW_UPDATE split (#454)

      *  Define STREAM_ID_NEEDED frame (#455)

   o  A NEW_CONNECTION_ID frame supports connection migration without
      linkability (#232, #491, #496)

   o  Transport parameters for 0-RTT are retained from a previous
      connection (#512)

      *  A client in 0-RTT no longer required to reset excess streams
         (#425, #479)

   o  Expanded security considerations (#440, #444, #445, #448)

C.2.  Since draft-ietf-quic-transport-01

   o  Defined short and long packet headers (#40, #148, #361)

   o  Defined a versioning scheme and stable fields (#51, #361)

   o  Define reserved version values for "greasing" negotiation (#112,
      #278)

   o  The initial packet number is randomized (#35, #283)
   o  Narrow the packet number encoding range requirement (#67, #286,
      #299, #323, #356)

   o  Defined client address validation (#52, #118, #120, #275)

   o  Define transport parameters as a TLS extension (#122) (#49, #122)

   o  SCUP and COPT parameters are no longer valid (#116, #117)

   o  Transport parameters for 0-RTT are either remembered from before,
      or assume default values (#126)

   o  The server chooses connection IDs in its final flight (#119, #349,
      #361)

   o  The server echoes the Connection ID and packet number fields when
      sending a Version Negotiation packet (#133, #295, #244)

   o  Definied  Defined a minimum packet size for the initial handshake packet
      from the client (#69, #136, #139, #164)

   o  Path MTU Discovery (#64, #106)

   o  The initial handshake packet from the client needs to fit in a
      single packet (#338)

   o  Forbid acknowledgment of packets containing only ACK and PADDING
      (#291)

   o  Require that frames are processed when packets are acknowledged
      (#381, #341)

   o  Removed the STOP_WAITING frame (#66)

   o  Don't require retransmission of old timestamps for lost ACK frames
      (#308)

   o  Clarified that frames are not retransmitted, but the information
      in them can be (#157, #298)

   o  Error handling definitions (#335)

   o  Split error codes into four sections (#74)

   o  Forbid the use of Public Reset where CONNECTION_CLOSE is possible
      (#289)

   o  Define packet protection rules (#336)
   o  Require that stream be entirely delivered or reset, including
      acknowledgment of all STREAM frames or the RST_STREAM, before it
      closes (#381)

   o  Remove stream reservation from state machine (#174, #280)

   o  Only stream 0 1 does not contributing contribute to connection-level flow control
      (#204)

   o  Stream 1 counts towards the maximum concurrent stream limit (#201,
      #282)

   o  Remove connection-level flow control exclusion for some streams
      (except 1) (#246)

   o  RST_STREAM affects connection-level flow control (#162, #163)

   o  Flow control accounting uses the maximum data offset on each
      stream, rather than bytes received (#378)

   o  Moved length-determining fields to the start of STREAM and ACK
      (#168, #277)

   o  Added the ability to pad between frames (#158, #276)

   o  Remove error code and reason phrase from GOAWAY (#352, #355)

   o  GOAWAY includes a final stream number for both directions (#347)

   o  Error codes for RST_STREAM and CONNECTION_CLOSE are now at a
      consistent offset (#249)

   o  Defined priority as the responsibility of the application protocol
      (#104, #303)

C.2.

C.3.  Since draft-ietf-quic-transport-00: draft-ietf-quic-transport-00

   o  Replaced DIVERSIFICATION_NONCE flag with KEY_PHASE flag

   o  Defined versioning

   o  Reworked description of packet and frame layout

   o  Error code space is divided into regions for each component

   o  Use big endian for all numeric values

C.3.

C.4.  Since draft-hamilton-quic-transport-protocol-01: draft-hamilton-quic-transport-protocol-01

   o  Adopted as base for draft-ietf-quic-tls. draft-ietf-quic-tls

   o  Updated authors/editors list. list

   o  Added IANA Considerations section. section

   o  Moved Contributors and Acknowledgments to appendices. appendices

Authors' Addresses

   Jana Iyengar (editor)
   Google

   Email: jri@google.com

   Martin Thomson (editor)
   Mozilla

   Email: martin.thomson@gmail.com