tcp/connection.rs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
use bytes::{Buf, Bytes};
use std::io::{Read, Write};
use std::net::SocketAddrV4;
use crate::buffer::{RecvQueue, Segment};
use crate::seq::{Seq, SeqRange};
use crate::util::time::Instant;
use crate::window_scaling::WindowScaling;
use crate::{
Ipv4Header, Payload, PopPacketError, PushPacketError, RecvError, SendError, TcpConfig,
TcpFlags, TcpHeader,
};
/// Information for a TCP connection. Equivalent to the Transmission Control Block (TCB).
#[derive(Debug)]
pub(crate) struct Connection<I: Instant> {
pub(crate) config: TcpConfig,
pub(crate) local_addr: SocketAddrV4,
pub(crate) remote_addr: SocketAddrV4,
pub(crate) send: ConnectionSend<I>,
pub(crate) recv: Option<ConnectionRecv>,
pub(crate) need_to_ack: bool,
pub(crate) last_advertised_window: Option<u32>,
pub(crate) window_scaling: WindowScaling,
pub(crate) send_rst_if_recv_payload: bool,
pub(crate) is_reset: bool,
pub(crate) need_to_send_rst: bool,
}
impl<I: Instant> Connection<I> {
/// The max number of bytes allowed in the send and receive buffers. These should be made
/// dynamic in the future.
const SEND_BUF_MAX: usize = 100_000;
const RECV_BUF_MAX: u32 = 100_000;
pub fn new(
local_addr: SocketAddrV4,
remote_addr: SocketAddrV4,
send_initial_seq: Seq,
config: TcpConfig,
) -> Self {
let mut rv = Self {
config,
local_addr,
remote_addr,
send: ConnectionSend::new(send_initial_seq),
recv: None,
need_to_ack: true,
last_advertised_window: None,
window_scaling: WindowScaling::new(),
send_rst_if_recv_payload: false,
is_reset: false,
need_to_send_rst: false,
};
// disable window scaling if it's disabled in the config
if !rv.config.window_scaling_enabled {
rv.window_scaling.disable();
}
rv
}
pub fn into_recv_buffer(self) -> Option<RecvQueue> {
if let Some(recv) = self.recv {
return Some(recv.buffer);
}
None
}
/// Returns `true` if the packet header src/dst addresses match this connection.
pub fn packet_addrs_match(&self, header: &TcpHeader) -> bool {
header.src() == self.remote_addr && header.dst() == self.local_addr
}
pub fn send_fin(&mut self) {
self.send.buffer.add_fin();
self.send.is_closed = true;
}
pub fn send_rst(&mut self) {
self.need_to_send_rst = true;
self.is_reset = true;
}
/// If any new payload bytes are received, the connection will be reset.
pub fn send_rst_if_recv_payload(&mut self) {
self.send_rst_if_recv_payload = true;
}
pub fn send(&mut self, reader: impl Read, len: usize) -> Result<usize, SendError> {
// if the buffer is full
if !self.send_buf_has_space() {
return Err(SendError::Full);
}
let send_buffer_len = self.send.buffer.len() as usize;
let send_buffer_space = Self::SEND_BUF_MAX.saturating_sub(send_buffer_len);
let len = std::cmp::min(len, send_buffer_space);
if let Err(e) = self.send.buffer.add_data(reader, len) {
return Err(SendError::Io(e));
}
Ok(len)
}
pub fn recv(&mut self, writer: impl Write, len: usize) -> Result<usize, RecvError> {
let recv = self.recv.as_mut().unwrap();
if recv.buffer.is_empty() {
return Err(RecvError::Empty);
}
recv.buffer.read(writer, len).map_err(RecvError::Io)
}
pub fn push_packet(
&mut self,
header: &TcpHeader,
payload: Payload,
) -> Result<u32, PushPacketError> {
if self.is_reset {
panic!(
"The connection has already been reset, so why are we being given more packets?"
);
}
// process RST packets
if header.flags.contains(TcpFlags::RST) {
let seq = Seq::new(header.seq);
let recv_window = self.recv_window();
// TODO: figure out how to properly handle weird RST packets (for example RST packets
// with payload data)
let Some(recv_window) = recv_window else {
// we haven't received a SYN yet, so we'll trust the RST
self.is_reset = true;
return Ok(0);
};
// RFC 9293 3.10.7.4.:
// > If the RCV.WND is zero, no segments will be acceptable, but special allowance
// > should be made to accept valid ACKs, URGs, and RSTs.
if seq == recv_window.start {
// RFC 9293 3.10.7.4.:
// > If the RST bit is set and the sequence number exactly matches the next expected
// > sequence number (RCV.NXT), then TCP endpoints MUST reset the connection in the
// > manner prescribed below according to the connection state.
self.is_reset = true;
return Ok(0);
}
if recv_window.contains(seq) {
// RFC 9293 3.10.7.4.:
// > If the RST bit is set and the sequence number does not exactly match the next
// > expected sequence value, yet is within the current receive window, TCP
// > endpoints MUST send an acknowledgment (challenge ACK):
// >
// > <SEQ=SND.NXT><ACK=RCV.NXT><CTL=ACK>
// >
// > After sending the challenge ACK, TCP endpoints MUST drop the unacceptable
// > segment and stop processing the incoming packet further.
// TODO: Setting `need_to_ack` to true isn't enough to send an acknowledgement
// exactly as described above, since the next `pop_packet` may next try to
// retransmit something which would have a different sequence number. But not sure
// if this really matters in practice since the peer would receive the packet and
// send another RST packet based on the ACK value we send.
self.need_to_ack = true;
return Ok(0);
}
// RFC 9293 3.10.7.4.:
// > If the RST bit is set and the sequence number is outside the current receive
// > window, silently drop the segment.
return Ok(0);
}
// process the first SYN packet
if self.recv.is_none() && header.flags.contains(TcpFlags::SYN) {
// we needed to know the sender's initial sequence number before we could initialize the
// receiving part of the connection
let seq = Seq::new(header.seq);
self.recv = Some(ConnectionRecv::new(seq));
self.window_scaling.received_syn(header.window_scale);
}
// We need to keep track of if the original packet had the SYN flag set, even if we trim a
// old retransmitted SYN flag from the packet below. If it was sent with a SYN flag, then we
// must not apply the window scale to the window size in the packet, even if the SYN's
// sequence number isn't within the receive window.
//
// RFC 7323 2.2.:
// > The window field in a segment where the SYN bit is set (i.e., a <SYN> or <SYN,ACK>)
// > MUST NOT be scaled.
//
// TODO: be careful about this if we support a reassembly queue in the future
let original_packet_had_syn = header.flags.contains(TcpFlags::SYN);
// trim the segment so that it only contains data/flags that fit within the receive window
let recv_window = self.recv_window().unwrap();
let Some((header, payload)) = trim_segment(header, payload, &recv_window) else {
// the sequence range of the segment does not overlap with the receive window, so we
// must drop the packet and send an ACK
self.need_to_ack = true;
return Ok(0);
};
let Some(recv) = self.recv.as_mut() else {
// we received a non-SYN packet before the first SYN packet
self.send_rst();
return Ok(0);
};
// if we've been told to send a RST when we receive new payload data, and we did receive new
// payload data
if self.send_rst_if_recv_payload && !payload.is_empty() {
self.send_rst();
return Ok(0);
}
// if we've previously received a FIN packet, and now we've received a payload/SYN/FIN
// packet that is within the receive window
if recv.is_closed
&& (!payload.is_empty() || header.flags.intersects(TcpFlags::SYN | TcpFlags::FIN))
{
self.send_rst();
return Ok(0);
}
let mut pushed_len = 0;
// the receive buffer's initial next sequence number; useful so we can check if we need to
// acknowledge or not
let initial_seq = recv.buffer.next_seq();
if !recv.is_closed {
if header.flags.contains(TcpFlags::SYN) {
if recv.buffer.syn_added() {
// this is the second SYN we've received
// TODO: We can follow RFC 793 or RFC 5961 here. 793 is probably easiest, and we
// should send an RST and move to the "closed" state.
self.send_rst();
return Ok(0);
}
recv.buffer.add_syn();
}
let syn_len = if header.flags.contains(TcpFlags::SYN) {
1
} else {
0
};
let payload_len = payload.len();
let payload_seq = (payload_len != 0).then_some(Seq::new(header.seq) + syn_len);
let fin_seq = header
.flags
.contains(TcpFlags::FIN)
.then_some(Seq::new(header.seq) + syn_len + payload_len);
if let Some(payload_seq) = payload_seq {
if payload_seq == recv.buffer.next_seq() {
pushed_len += payload.len();
for chunk in payload.0 {
recv.buffer.add(chunk);
}
} else {
// TODO: store (truncated?) out-of-order packet
}
}
if let Some(fin_seq) = fin_seq {
if fin_seq == recv.buffer.next_seq() {
recv.buffer.add_fin();
recv.is_closed = true;
} else {
// TODO: store (truncated?) out of order packet
}
}
}
// we've added to the receive buffer (payload, syn, or fin), so we need to send an
// acknowledgement
if recv.buffer.next_seq() != initial_seq {
self.need_to_ack = true;
}
// update the send window, applying the window scale shift only if it wasn't a SYN packet
// TODO: should we still update the window if the ACK was not in the valid range?
if original_packet_had_syn {
self.send.window = u32::from(header.window_size);
} else {
self.send.window =
u32::from(header.window_size) << self.window_scaling.send_window_scale_shift();
}
if header.flags.contains(TcpFlags::ACK) {
let valid_ack_range = SeqRange::new(
self.send.buffer.start_seq() + 1,
self.send.buffer.next_seq() + 1,
);
if valid_ack_range.contains(Seq::new(header.ack)) {
// the SYN is always first, so if a new sequence number has been acknowledged, then
// either it's acknowledging the SYN, or the SYN has been acknowledged in the past
if Seq::new(header.ack) != self.send.buffer.start_seq() {
self.send.syn_acked = true;
}
self.send.buffer.advance_start(Seq::new(header.ack));
}
}
Ok(pushed_len)
}
pub fn pop_packet(&mut self, now: I) -> Result<(TcpHeader, Payload), PopPacketError> {
let (seq_range, mut flags, payload) =
self.next_segment().ok_or(PopPacketError::NoPacket)?;
// After this point we must always send a packet. If we don't, then we're not being
// consistent with `self.wants_to_send()`.
debug_assert!(self.wants_to_send());
let header_ack = if let Some(recv) = self.recv.as_ref() {
// we've received a SYN packet (either now or in the past), so should always acknowledge
flags.insert(TcpFlags::ACK);
recv.buffer.next_seq()
} else {
// not setting the ACK flag, so this can probably be anything
Seq::new(0)
};
let header_window_size;
let header_window_scale;
if flags.contains(TcpFlags::SYN) {
if self.window_scaling.can_send_window_scale() {
// The receive buffer capacity at the time the SYN is sent decides the window
// scaling to use. This effectively limits future receive buffer capacity increases
// since the receive window will forever have a ceiling set here by the window
// scale.
let shift = WindowScaling::scale_shift_for_max_window(self.recv_buffer_capacity());
header_window_scale = Some(shift);
} else {
header_window_scale = None;
}
// don't actually apply this window scale in the SYN packet
//
// RFC 7323 2.2.:
// > The window field in a segment where the SYN bit is set (i.e., a <SYN> or <SYN,ACK>)
// > MUST NOT be scaled.
header_window_size = self.recv_window_len();
self.last_advertised_window = Some(header_window_size);
// Make sure we're sending a valid 2-byte window size. We haven't called
// `WindowScaling::sent_syn()` yet, so `Self::recv_window_len()` should not have
// returned a window size larger than `u16::MAX`.
debug_assert!(header_window_size <= u16::MAX as u32);
self.window_scaling.sent_syn(header_window_scale);
} else {
// don't send a window scale
//
// RFC 7323 2.1.:
// > The exponent of the scale factor is carried in a TCP option, Window Scale. This
// > option is sent only in a <SYN> segment (a segment with the SYN bit on), [...]
header_window_scale = None;
let shift = self.window_scaling.recv_window_scale_shift();
header_window_size = self.recv_window_len() >> shift;
// this is the value the peer will see (precision is intentionally lost due to bit-shift)
self.last_advertised_window = Some(header_window_size << shift);
}
let header = TcpHeader {
ip: Ipv4Header {
src: *self.local_addr.ip(),
dst: *self.remote_addr.ip(),
},
flags,
src_port: self.local_addr.port(),
dst_port: self.remote_addr.port(),
seq: seq_range.start.into(),
ack: header_ack.into(),
window_size: header_window_size.try_into().unwrap(),
selective_acks: None,
window_scale: header_window_scale,
timestamp: None,
timestamp_echo: None,
};
// we're sending the most up-to-date acknowledgement
self.need_to_ack = false;
// inform the buffer that we transmitted this segment
self.send.buffer.mark_as_transmitted(seq_range.end, now);
if header.flags.contains(TcpFlags::RST) {
assert!(self.need_to_send_rst);
self.need_to_send_rst = false;
}
Ok((header, payload))
}
/// Returns a segment that is ready to send. This may be a data segment (a segment containing a
/// SYN/FIN flag and/or payload data), a RST segment, or an empty segment. Even if this returns
/// an empty segment, it must be sent with the correct acknowledgement number, window size, etc
/// as it may represent an acknowledgement or window update.
fn next_segment(&self) -> Option<(SeqRange, TcpFlags, Payload)> {
// should be inlined
self._next_segment()
}
/// Returns true if ready to send a packet.
pub fn wants_to_send(&self) -> bool {
// should be inlined
self._next_segment().is_some()
}
/// Do not call directly. Use either `next_segment()` or `wants_to_send()`.
///
/// Since `wants_to_send()` is only interested in whether the result is `Some`, by inlining this
/// function the compiler should hopefully optimize it to remove unnecessary values that will be
/// immediately discarded. I'm uncertain whether there's really much that can be optimized here
/// though, but splitting it into functions will at least help us notice if either function is
/// showing up in a profile/heatmap. Since the function is large and is `inline(always)` we only
/// call it from two functions, `next_segment()` and `wants_to_send()`.
#[inline(always)]
fn _next_segment(&self) -> Option<(SeqRange, TcpFlags, Payload)> {
if self.need_to_send_rst {
let seq = self
.send
.buffer
.next_not_transmitted(0)
.map(|x| x.0)
.unwrap_or(self.send.buffer.next_seq());
let seq_range = SeqRange::new(seq, seq);
return Some((seq_range, TcpFlags::RST, Payload::default()));
}
// if the connection has been reset and we don't need to send a RST packet, never send any
// future packets
if self.is_reset {
return None;
}
let (seq_range, syn_fin_flags, payload) = 'packet: {
// if we have syn/fin/payload data to send
if let Some((seq_range, syn_fin_flags, payload)) = self.next_data_segment() {
break 'packet (seq_range, syn_fin_flags, payload);
}
let mut send_empty_packet = false;
// do we need to send an acknowledgement?
if self.need_to_ack {
send_empty_packet = true;
}
// do we need to send a window update?
if let Some(window) = self.recv_window().map(|x| x.len()) {
let window_scale = self.window_scaling.recv_window_scale_shift();
let apparent_window = window >> window_scale << window_scale;
if self.last_advertised_window != Some(apparent_window) {
send_empty_packet = true;
}
}
if send_empty_packet {
// use the sequence number of the next unsent message if we have one buffered,
// otherwise get the next sequence number from the buffer
let seq = self
.send
.buffer
.next_not_transmitted(0)
.map(|x| x.0)
.unwrap_or(self.send.buffer.next_seq());
let seq_range = SeqRange::new(seq, seq);
break 'packet (seq_range, TcpFlags::empty(), Payload::default());
}
return None;
};
// if not sending a SYN packet and window scaling isn't yet confirmed
if !syn_fin_flags.contains(TcpFlags::SYN) && !self.window_scaling.is_configured() {
// we cannot send a non-SYN packet since non-SYN packets must apply window scaling, but
// we haven't yet confirmed if we're using window scaling or not
return None;
}
Some((seq_range, syn_fin_flags, payload))
}
/// Returns a data segment that is ready to send. This is a segment containing a SYN/FIN flag
/// and/or payload data. Even if this returns `None`, we may still want to send some other
/// segment such as an acknowledgement or window update (see `Self::next_segment`).
fn next_data_segment(&self) -> Option<(SeqRange, TcpFlags, Payload)> {
let send_window = self.send_window();
let mut chunks = Vec::new();
let mut syn_fin_flags = TcpFlags::empty();
let mut seq_start = None;
let mut seq_len = 0;
let mut payload_bytes_len = 0;
// roughly represents the MSS
// TODO: handle the MSS properly
const MAX_BYTES_PER_PACKET: u32 = 1500;
// do we have syn/fin/payload data to send?
while let Some((seq, segment)) = self.send.buffer.next_not_transmitted(seq_len) {
// if no bytes of this segment fit within the send window
if !send_window.contains(seq) {
break;
}
// if we can't send any more payload bytes
if payload_bytes_len == MAX_BYTES_PER_PACKET {
break;
}
// if this is the first returned segment, keep track of the start
if seq_start.is_none() {
seq_start = Some(seq);
}
match segment {
Segment::Syn => {
syn_fin_flags.insert(TcpFlags::SYN);
seq_len += segment.len();
}
Segment::Fin => {
syn_fin_flags.insert(TcpFlags::FIN);
seq_len += segment.len();
}
Segment::Data(mut chunk) => {
let allowed_payload_len =
MAX_BYTES_PER_PACKET.saturating_sub(payload_bytes_len);
let allowed_seq_len = send_window.end - seq;
let allowed_len = std::cmp::min(allowed_payload_len, allowed_seq_len);
chunk.truncate(std::cmp::min(chunk.len(), allowed_len.try_into().unwrap()));
let chunk_len: u32 = chunk.len().try_into().unwrap();
payload_bytes_len += chunk_len;
seq_len += chunk_len;
chunks.push(chunk);
}
};
// we shouldn't be sending more than allowed
debug_assert!(payload_bytes_len <= MAX_BYTES_PER_PACKET);
}
if !chunks.is_empty() || !syn_fin_flags.is_empty() {
let seq_start = seq_start.unwrap();
let seq_range = SeqRange::new(seq_start, seq_start + seq_len);
return Some((seq_range, syn_fin_flags, Payload(chunks)));
}
None
}
/// Returns true if we received a RST packet, or if we want to send a RST packet.
pub fn is_reset(&self) -> bool {
self.is_reset
}
/// Returns true if we received a SYN packet from the peer.
pub fn received_syn(&self) -> bool {
// we don't construct the receive part of the connection until we've received the SYN
self.recv.is_some()
}
/// Returns true if we received a FIN packet from the peer.
pub fn received_fin(&self) -> bool {
self.recv.as_ref().map(|x| x.is_closed).unwrap_or(false)
}
/// Returns true if the peer acknowledged the SYN packet we sent.
pub fn syn_was_acked(&self) -> bool {
self.send.syn_acked
}
/// Returns true if the peer acknowledged the FIN packet we sent.
pub fn fin_was_acked(&self) -> bool {
self.send.is_closed && self.send.buffer.start_seq() == self.send.buffer.next_seq()
}
/// Returns true if the send buffer has space available. Does not consider whether the
/// connection is open/closed, either due to FIN packets or `shutdown()`.
pub fn send_buf_has_space(&self) -> bool {
let send_buffer_len = self.send.buffer.len() as usize;
send_buffer_len < Self::SEND_BUF_MAX
}
/// Returns true if the recv buffer has data to read. Does not consider whether the connection
/// is open/closed, either due to FIN packets or `shutdown()`.
pub fn recv_buf_has_data(&self) -> bool {
let is_empty = self
.recv
.as_ref()
.map(|x| x.buffer.is_empty())
.unwrap_or(true);
!is_empty
}
pub(crate) fn send_window(&self) -> SeqRange {
// the buffer stores unsent/unacked data, so the buffer starts at the lowest unacked
// sequence number
let window_left = self.send.buffer.start_seq();
SeqRange::new(window_left, window_left + self.send.window)
}
/// Returns the size of the receive window. This is useful when we only need the size of the
/// window and we may not have received the SYN packet yet, so cannot construct the range.
pub(crate) fn recv_window_len(&self) -> u32 {
if let Some(recv_window) = self.recv_window() {
return recv_window.len();
}
let window_max = self.window_scaling.recv_window_max();
std::cmp::min(self.recv_buffer_capacity(), window_max)
}
/// Returns the receive window if we've received a SYN packet.
pub(crate) fn recv_window(&self) -> Option<SeqRange> {
let recv = self.recv.as_ref()?;
let window_left = recv.buffer.next_seq();
let window_max = self.window_scaling.recv_window_max();
let window_len = self
.recv_buffer_capacity()
.saturating_sub(recv.buffer.len());
let window_len = std::cmp::min(window_len, window_max);
Some(SeqRange::new(window_left, window_left + window_len))
}
/// The total capacity of the receive buffer.
fn recv_buffer_capacity(&self) -> u32 {
Self::RECV_BUF_MAX
}
}
#[derive(Debug)]
pub(crate) struct ConnectionSend<I: Instant> {
pub(crate) buffer: super::buffer::SendQueue<I>,
pub(crate) window: u32,
pub(crate) is_closed: bool,
pub(crate) syn_acked: bool,
}
impl<I: Instant> ConnectionSend<I> {
pub fn new(initial_seq: Seq) -> Self {
Self {
buffer: super::buffer::SendQueue::new(initial_seq),
// we don't know the peer's receive window, so choose something conservative
window: 2048,
is_closed: false,
syn_acked: false,
}
}
}
#[derive(Debug)]
pub(crate) struct ConnectionRecv {
pub(crate) buffer: super::buffer::RecvQueue,
pub(crate) is_closed: bool,
}
impl ConnectionRecv {
pub fn new(initial_seq: Seq) -> Self {
Self {
buffer: super::buffer::RecvQueue::new(initial_seq),
is_closed: false,
}
}
}
/// Trims the segment `header` and `payload` such that only bytes in the sequence `range` remain.
/// This may modify the segment sequence number, SYN/FIN flags, or payload.
fn trim_segment(
header: &TcpHeader,
payload: Payload,
range: &SeqRange,
) -> Option<(TcpHeader, Payload)> {
let seq = Seq::new(header.seq);
let syn_len = if header.flags.contains(TcpFlags::SYN) {
1
} else {
0
};
let fin_len = if header.flags.contains(TcpFlags::FIN) {
1
} else {
0
};
let payload_len = payload.len();
let header_range = SeqRange::new(seq, seq + syn_len + payload_len + fin_len);
let intersection = header_range.intersection(range)?;
if intersection == header_range {
// in the common case where the segment is completely contained within the range, return
// early without any modifications
return Some((*header, payload));
}
let include_syn = syn_len == 1 && range.contains(header_range.start);
let include_fin = fin_len == 1 && range.contains(header_range.end - 1);
let payload_seq = seq + syn_len;
let new_payload = match trim_payload(payload_seq, payload, range) {
Some((new_seq, new_payload)) => {
assert_eq!(
new_seq,
intersection.start + if include_syn { 1 } else { 0 }
);
new_payload
}
None => Payload::default(),
};
let mut new_flags = header.flags;
new_flags.set(TcpFlags::SYN, include_syn);
new_flags.set(TcpFlags::FIN, include_fin);
let new_header = TcpHeader {
seq: intersection.start.into(),
flags: new_flags,
..*header
};
Some((new_header, new_payload))
}
/// Trims `payload`, which starts at a given `seq` number, such that only bytes in the sequence
/// `range` remain.
///
/// If the two ranges do not intersect `None` will be returned. A `None` is also returned if the
/// range intersects the payload twice, for example if the payload covers the range 100..200 and the
/// given range covers 180..120, but this shouldn't occur for reasonable TCP sequence number ranges.
/// The returned payload may be empty if the original `payload` was empty or the `range` was empty,
/// but they still intersect according to [`SeqRange::intersection`].
fn trim_payload(seq: Seq, mut payload: Payload, range: &SeqRange) -> Option<(Seq, Payload)> {
let payload_range = SeqRange::new(seq, seq + payload.len());
let intersection = payload_range.intersection(range)?;
if payload_range == intersection {
// in the common case where the payload is completely contained within the range, return
// early without any modifications
return Some((seq, payload));
}
// the sequence number of the current chunk
let mut seq_cursor = seq;
// we could use `retain` here and remove empty/out-of-bounds chunks, but it's simpler and
// probably faster to avoid shifting elements around and just leave empty chunks (and replace
// out-of-bounds chunks with empty chunks)
for chunk in &mut payload.0 {
let original_chunk_len = chunk.len().try_into().unwrap();
// `take` will replace the current chunk with an empty chunk
if let Some((_seq, new_chunk)) = trim_chunk(seq_cursor, std::mem::take(chunk), range) {
*chunk = new_chunk;
}
seq_cursor += original_chunk_len;
}
debug_assert_eq!(payload.len(), intersection.len());
Some((intersection.start, payload))
}
/// Trims `chunk`, which starts at a given `seq` number, such that only bytes in the sequence
/// `range` remain.
///
/// If the two ranges do not intersect `None` will be returned. A `None` is also returned if the
/// range intersects the chunk twice, for example if the chunk covers the range 100..200 and the
/// given range covers 180..120, but this shouldn't occur for reasonable TCP sequence number ranges.
/// The returned chunk may be empty if the original `chunk` was empty or the `range` was empty, but
/// they still intersect according to [`SeqRange::intersection`].
fn trim_chunk(seq: Seq, mut chunk: Bytes, range: &SeqRange) -> Option<(Seq, Bytes)> {
let chunk_range = SeqRange::new(seq, seq + chunk.len().try_into().unwrap());
let intersection = chunk_range.intersection(range)?;
let new_offset = intersection.start - seq;
let new_len = intersection.len();
let new_offset: usize = new_offset.try_into().unwrap();
let new_len: usize = new_len.try_into().unwrap();
// update the existing `Bytes` object rather than using `slice()` to avoid an atomic operation
chunk.advance(new_offset);
chunk.truncate(new_len);
Some((intersection.start, chunk))
}
#[cfg(test)]
mod tests {
use super::*;
use std::net::Ipv4Addr;
// helper to make the tests fit on a single line
fn range(start: u32, end: u32) -> SeqRange {
SeqRange::new(Seq::new(start), Seq::new(end))
}
// helper to make the tests fit on a single line
fn seq(val: u32) -> Seq {
Seq::new(val)
}
// helper to make the tests fit on a single line
fn bytes<const N: usize>(x: &[u8; N]) -> Bytes {
Box::<[u8]>::from(x.as_slice()).into()
}
// helper to make the tests fit on a single line
macro_rules! payload {
() => {
Payload([].into())
};
($($slices:literal),+) => {
{
let iter = ([$(&$slices[..]),+]).into_iter().map(|x| Bytes::copy_from_slice(&x));
Payload(iter.collect())
}
};
}
#[test]
fn test_trim_segment() {
fn test_trim(
flags: TcpFlags,
seq: Seq,
payload: impl Into<Payload>,
range: SeqRange,
) -> Option<(TcpFlags, Seq, Bytes)> {
let header = TcpHeader {
ip: Ipv4Header {
src: Ipv4Addr::UNSPECIFIED,
dst: Ipv4Addr::UNSPECIFIED,
},
flags,
src_port: 0,
dst_port: 0,
seq: seq.into(),
ack: 0,
window_size: 0,
selective_acks: None,
window_scale: None,
timestamp: None,
timestamp_echo: None,
};
let (header, payload) = trim_segment(&header, payload.into(), &range)?;
let payload = payload.concat();
Some((header.flags, Seq::new(header.seq), payload))
}
const SYN: TcpFlags = TcpFlags::SYN;
const FIN: TcpFlags = TcpFlags::FIN;
const ACK: TcpFlags = TcpFlags::ACK;
const EMPTY: TcpFlags = TcpFlags::empty();
assert_eq!(test_trim(EMPTY, seq(0), bytes(b""), range(1, 1)), None);
assert_eq!(test_trim(EMPTY, seq(1), bytes(b""), range(0, 1)), None);
assert_eq!(
test_trim(EMPTY, seq(0), bytes(b""), range(0, 0)),
Some((EMPTY, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(ACK, seq(0), bytes(b""), range(0, 0)),
Some((ACK, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(EMPTY, seq(0), bytes(b"123"), range(0, 0)),
Some((EMPTY, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(SYN, seq(0), bytes(b""), range(0, 0)),
Some((EMPTY, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(FIN, seq(0), bytes(b""), range(0, 0)),
Some((EMPTY, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(EMPTY, seq(0), bytes(b""), range(0, 2)),
Some((EMPTY, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(EMPTY, seq(0), bytes(b"123"), range(0, 2)),
Some((EMPTY, seq(0), bytes(b"12"))),
);
assert_eq!(
test_trim(SYN, seq(0), bytes(b"123"), range(0, 2)),
Some((SYN, seq(0), bytes(b"1"))),
);
assert_eq!(
test_trim(SYN | FIN, seq(0), bytes(b"123"), range(0, 2)),
Some((SYN, seq(0), bytes(b"1"))),
);
assert_eq!(
test_trim(SYN | FIN, seq(0), bytes(b"123"), range(1, 2)),
Some((EMPTY, seq(1), bytes(b"1"))),
);
assert_eq!(
test_trim(SYN | FIN, seq(0), bytes(b"123"), range(1, 5)),
Some((FIN, seq(1), bytes(b"123"))),
);
assert_eq!(
test_trim(SYN | FIN, seq(0), bytes(b"123"), range(0, 1)),
Some((SYN, seq(0), bytes(b""))),
);
assert_eq!(
test_trim(SYN | FIN, seq(4), bytes(b"123"), range(0, 5)),
Some((SYN, seq(4), bytes(b""))),
);
assert_eq!(
test_trim(SYN | FIN | ACK, seq(3), bytes(b"123"), range(0, 5)),
Some((SYN | ACK, seq(3), bytes(b"1"))),
);
}
#[test]
fn test_trim_payload() {
fn test_trim(seq: Seq, payload: Payload, range: SeqRange) -> Option<(Seq, Vec<Bytes>)> {
let (seq, payload) = trim_payload(seq, payload, &range)?;
Some((seq, payload.0))
}
assert_eq!(
test_trim(seq(0), payload![b""], range(0, 0)),
Some((seq(0), payload![b""].0)),
);
assert_eq!(
test_trim(seq(0), payload![], range(0, 0)),
Some((seq(0), vec![])),
);
assert_eq!(test_trim(seq(1), payload![b""], range(0, 0)), None);
assert_eq!(test_trim(seq(1), payload![b""], range(0, 1)), None);
assert_eq!(
test_trim(seq(1), payload![b""], range(0, 2)),
Some((seq(1), payload![b""].0)),
);
assert_eq!(
test_trim(seq(0), payload![b"a"], range(0, 0)),
Some((seq(0), payload![b""].0)),
);
assert_eq!(
test_trim(seq(0), payload![b"a"], range(0, 1)),
Some((seq(0), payload![b"a"].0)),
);
assert_eq!(
test_trim(seq(0), payload![b"ab"], range(0, 1)),
Some((seq(0), payload![b"a"].0)),
);
assert_eq!(
test_trim(seq(0), payload![b"abcdefg"], range(2, 4)),
Some((seq(2), payload![b"cd"].0)),
);
assert_eq!(
test_trim(seq(3), payload![b"abcdefg"], range(2, 4)),
Some((seq(3), payload![b"a"].0)),
);
assert_eq!(
test_trim(seq(3), payload![b"abcdefg"], range(2, 20)),
Some((seq(3), payload![b"abcdefg"].0)),
);
assert_eq!(
test_trim(seq(3), payload![b"abcdefg"], range(9, 20)),
Some((seq(9), payload![b"g"].0)),
);
assert_eq!(test_trim(seq(3), payload![b"abcdefg"], range(10, 20)), None);
// second test intersects twice, so returns `None`
assert_eq!(
test_trim(seq(5), payload![b"abcdefg"], range(8, 5)),
Some((seq(8), payload![b"defg"].0)),
);
assert_eq!(test_trim(seq(5), payload![b"abcdefg"], range(8, 7)), None);
// cut off right edge
assert_eq!(
test_trim(seq(0), payload![b"a", b"bcd"], range(0, 3)),
Some((seq(0), payload![b"a", b"bc"].0)),
);
// cut off left edge
assert_eq!(
test_trim(seq(0), payload![b"abc", b"d"], range(1, 4)),
Some((seq(1), payload![b"bc", b"d"].0)),
);
// cut off left and right edge
assert_eq!(
test_trim(seq(0), payload![b"abc", b"def", b"ghi"], range(1, 8)),
Some((seq(1), payload![b"bc", b"def", b"gh"].0)),
);
// cut off left and right chunks
assert_eq!(
test_trim(seq(0), payload![b"abc", b"def", b"ghi"], range(3, 6)),
Some((seq(3), payload![b"", b"def", b""].0)),
);
// cut off left and right edges of same chunk
assert_eq!(
test_trim(seq(0), payload![b"abc", b"def", b"ghi"], range(4, 5)),
Some((seq(4), payload![b"", b"e", b""].0)),
);
assert_eq!(
test_trim(
seq(0),
payload![b"", b"abc", b"", b"de", b"", b"f", b"", b"ghi"],
range(3, 6)
),
Some((
seq(3),
payload![b"", b"", b"", b"de", b"", b"f", b"", b""].0
)),
);
}
#[test]
fn test_trim_chunk() {
fn test_trim(seq: Seq, chunk: Bytes, range: SeqRange) -> Option<(Seq, Bytes)> {
trim_chunk(seq, chunk, &range)
}
assert_eq!(
test_trim(seq(0), bytes(b""), range(0, 0)),
Some((seq(0), bytes(b""))),
);
assert_eq!(test_trim(seq(1), bytes(b""), range(0, 0)), None);
assert_eq!(test_trim(seq(1), bytes(b""), range(0, 1)), None);
assert_eq!(
test_trim(seq(1), bytes(b""), range(0, 2)),
Some((seq(1), bytes(b""))),
);
assert_eq!(
test_trim(seq(0), bytes(b"a"), range(0, 0)),
Some((seq(0), bytes(b""))),
);
assert_eq!(
test_trim(seq(0), bytes(b"a"), range(0, 1)),
Some((seq(0), bytes(b"a"))),
);
assert_eq!(
test_trim(seq(0), bytes(b"ab"), range(0, 1)),
Some((seq(0), bytes(b"a"))),
);
assert_eq!(
test_trim(seq(0), bytes(b"abcdefg"), range(2, 4)),
Some((seq(2), bytes(b"cd"))),
);
assert_eq!(
test_trim(seq(3), bytes(b"abcdefg"), range(2, 4)),
Some((seq(3), bytes(b"a"))),
);
assert_eq!(
test_trim(seq(3), bytes(b"abcdefg"), range(2, 20)),
Some((seq(3), bytes(b"abcdefg"))),
);
assert_eq!(
test_trim(seq(3), bytes(b"abcdefg"), range(9, 20)),
Some((seq(9), bytes(b"g"))),
);
assert_eq!(test_trim(seq(3), bytes(b"abcdefg"), range(10, 20)), None);
// second test intersects twice, so returns `None`
assert_eq!(
test_trim(seq(5), bytes(b"abcdefg"), range(8, 5)),
Some((seq(8), bytes(b"defg"))),
);
assert_eq!(test_trim(seq(5), bytes(b"abcdefg"), range(8, 7)), None);
}
}