[HICN-690] Transport Library Major Refactory

The current patch provides a major refactory of the transportlibrary.
A summary of the different components that underwent major modifications is
reported below.

- Transport protocol updates
The hierarchy of classes has been optimized to have common transport services
across different transport protocols. This can allow to customize a transport
protocol with new features.

- A new real-time communication protocol
The RTC protocol has been optimized in terms of algorithms to reduce
consumer-producer synchronization latency.

- A novel socket API
The API has been reworked to be easier to consumer but also to have a more
efficient integration in L4 proxies.

- Several performance improvements
A large number of performance improvements have been included in
particular to make the entire stack zero-copy and optimize cache miss.

- New memory buffer framework
Memory management has been reworked entirely to provide a more efficient infra
with a richer API. Buffers are now allocated in blocks and a single buffer
holds the memory for (1) the shared_ptr control block, (2) the metadata of the
packet (e.g. name, pointer to other buffers if buffer is chained and relevant
offsets), and (3) the packet itself, as it is sent/received over the network.

- A new slab allocator
Dynamic memory allocation is now managed by a novel slab allocator that is
optimised for packet processing and connection management. Memory is organized
in pools of blocks all of the same size which are used during the processing of
outgoing/incoming packets. When a memory block Is allocated is always taken
from a global pool and when it is deallocated is returned to the pool, thus
avoiding the cost of any heap allocation in the data path.

- New transport connectors
Consumer and producer end-points can communication either using an hicn packet
forwarder or with direct connector based on shared memories or sockets.
The usage of transport connectors typically for unit and funcitonal
testing but may have additional usage.

- Support for FEC/ECC for transport services
FEC/ECC via reed solomon is supported by default and made available to
transport services as a modular component. Reed solomon block codes is a
default FEC model that can be replaced in a modular way by many other
codes including RLNC not avaiable in this distribution.
The current FEC framework support variable size padding and efficiently
makes use of the infra memory buffers to avoid additiona copies.

- Secure transport framework for signature computation and verification
Crypto support is nativelty used in hICN for integrity and authenticity.
Novel support that includes RTC has been implemented and made modular
and reusable acrosso different transport protocols.

- TLS - Transport layer security over hicn
Point to point confidentiality is provided by integrating TLS on top of
hICN reliable and non-reliable transport. The integration is common and
makes a different use of the TLS record.

- MLS - Messaging layer security over hicn
MLS integration on top of hICN is made by using the MLSPP implemetation
open sourced by Cisco. We have included instrumentation tools to deploy
performance and functional tests of groups of end-points.

- Android support
The overall code has been heavily tested in Android environments and
has received heavy lifting to better run natively in recent Android OS.

Co-authored-by: Mauro Sardara <msardara@cisco.com>
Co-authored-by: Michele Papalini <micpapal@cisco.com>
Co-authored-by: Olivier Roques <oroques+fdio@cisco.com>
Co-authored-by: Giulio Grassi <gigrassi@cisco.com>
Change-Id: If477ba2fa686e6f47bdf96307ac60938766aef69
Signed-off-by: Luca Muscariello <muscariello@ieee.org>

1 files changed, 560 insertions, 0 deletions

diff --git a/libtransport/src/protocols/rtc/rtc_state.cc b/libtransport/src/protocols/rtc/rtc_state.cc
new file mode 100644
index 000000000..eabf8942c
--- /dev/null
+++ b/libtransport/src/protocols/rtc/rtc_state.cc
@@ -0,0 +1,560 @@
+/*
+ * Copyright (c) 2017-2021 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <protocols/rtc/rtc_consts.h>
+#include <protocols/rtc/rtc_state.h>
+
+namespace transport {
+
+namespace protocol {
+
+namespace rtc {
+
+RTCState::RTCState(ProbeHandler::SendProbeCallback &&rtt_probes_callback,
+                   DiscoveredRttCallback &&discovered_rtt_callback,
+                   asio::io_service &io_service)
+    : rtt_probes_(std::make_shared<ProbeHandler>(
+          std::move(rtt_probes_callback), io_service)),
+      discovered_rtt_callback_(std::move(discovered_rtt_callback)) {
+  init_rtt_timer_ = std::make_unique<asio::steady_timer>(io_service);
+  initParams();
+}
+
+RTCState::~RTCState() {}
+
+void RTCState::initParams() {
+  // packets counters (total)
+  sent_interests_ = 0;
+  sent_rtx_ = 0;
+  received_data_ = 0;
+  received_nacks_ = 0;
+  received_timeouts_ = 0;
+  received_probes_ = 0;
+
+  // loss counters
+  packets_lost_ = 0;
+  losses_recovered_ = 0;
+  first_seq_in_round_ = 0;
+  highest_seq_received_ = 0;
+  highest_seq_received_in_order_ = 0;
+  last_seq_nacked_ = 0;
+  loss_rate_ = 0.0;
+  residual_loss_rate_ = 0.0;
+
+  // bw counters
+  received_bytes_ = 0;
+  avg_packet_size_ = INIT_PACKET_SIZE;
+  production_rate_ = 0.0;
+  received_rate_ = 0.0;
+
+  // nack counter
+  nack_on_last_round_ = false;
+  received_nacks_last_round_ = 0;
+
+  // packets counter
+  received_packets_last_round_ = 0;
+  received_data_last_round_ = 0;
+  received_data_from_cache_ = 0;
+  data_from_cache_rate_ = 0;
+  sent_interests_last_round_ = 0;
+  sent_rtx_last_round_ = 0;
+
+  // round conunters
+  rounds_ = 0;
+  rounds_without_nacks_ = 0;
+  rounds_without_packets_ = 0;
+
+  last_production_seq_ = 0;
+  producer_is_active_ = false;
+  last_prod_update_ = 0;
+
+  // paths stats
+  path_table_.clear();
+  main_path_ = nullptr;
+
+  // packet received
+  received_or_lost_packets_.clear();
+
+  // pending interests
+  pending_interests_.clear();
+
+  // init rtt
+  first_interest_sent_ = ~0;
+  init_rtt_ = false;
+  rtt_probes_->setProbes(INIT_RTT_PROBE_INTERVAL, INIT_RTT_PROBES);
+  rtt_probes_->sendProbes();
+  setInitRttTimer(INIT_RTT_PROBE_RESTART);
+}
+
+// packet events
+void RTCState::onSendNewInterest(const core::Name *interest_name) {
+  uint64_t now = std::chrono::duration_cast<std::chrono::milliseconds>(
+                     std::chrono::steady_clock::now().time_since_epoch())
+                     .count();
+  uint32_t seq = interest_name->getSuffix();
+  pending_interests_.insert(std::pair<uint32_t, uint64_t>(seq, now));
+
+  if(sent_interests_ == 0) first_interest_sent_ = now;
+
+  sent_interests_++;
+  sent_interests_last_round_++;
+}
+
+void RTCState::onTimeout(uint32_t seq) {
+  auto it = pending_interests_.find(seq);
+  if (it != pending_interests_.end()) {
+    pending_interests_.erase(it);
+  }
+  received_timeouts_++;
+}
+
+void RTCState::onRetransmission(uint32_t seq) {
+  // remove the interest for the pendingInterest map only after the first rtx.
+  // in this way we can handle the ooo packets that come in late as normla
+  // packet. we consider a packet lost only if we sent at least an RTX for it.
+  // XXX this may become problematic if we stop the RTX transmissions
+  auto it = pending_interests_.find(seq);
+  if (it != pending_interests_.end()) {
+    pending_interests_.erase(it);
+    packets_lost_++;
+  }
+  sent_rtx_++;
+  sent_rtx_last_round_++;
+}
+
+void RTCState::onDataPacketReceived(const core::ContentObject &content_object,
+                                    bool compute_stats) {
+  uint32_t seq = content_object.getName().getSuffix();
+  if (compute_stats) {
+    updatePathStats(content_object, false);
+    received_data_last_round_++;
+  }
+  received_data_++;
+
+  struct data_packet_t *data_pkt =
+      (struct data_packet_t *)content_object.getPayload()->data();
+  uint64_t production_time = data_pkt->getTimestamp();
+  if (last_prod_update_ < production_time) {
+    last_prod_update_ = production_time;
+    uint32_t production_rate = data_pkt->getProductionRate();
+    production_rate_ = (double)production_rate;
+  }
+
+  updatePacketSize(content_object);
+  updateReceivedBytes(content_object);
+  addRecvOrLost(seq, PacketState::RECEIVED);
+
+  if (seq > highest_seq_received_) highest_seq_received_ = seq;
+
+  // the producer is responding
+  // it is generating valid data packets so we consider it active
+  producer_is_active_ = true;
+
+  received_packets_last_round_++;
+}
+
+void RTCState::onNackPacketReceived(const core::ContentObject &nack,
+                                    bool compute_stats) {
+  uint32_t seq = nack.getName().getSuffix();
+  struct nack_packet_t *nack_pkt =
+      (struct nack_packet_t *)nack.getPayload()->data();
+  uint64_t production_time = nack_pkt->getTimestamp();
+  uint32_t production_seq = nack_pkt->getProductionSegement();
+  uint32_t production_rate = nack_pkt->getProductionRate();
+
+  if (TRANSPORT_EXPECT_FALSE(main_path_ == nullptr) ||
+      last_prod_update_ < production_time) {
+    // update production rate
+    last_prod_update_ = production_time;
+    last_production_seq_ = production_seq;
+    production_rate_ = (double)production_rate;
+  }
+
+  if (compute_stats) {
+    // this is not an RTX
+    updatePathStats(nack, true);
+    nack_on_last_round_ = true;
+  }
+
+  // for statistics pourpose we log all nacks, also the one received for
+  // retransmitted packets
+  received_nacks_++;
+  received_nacks_last_round_++;
+
+  if (production_seq > seq) {
+    // old nack, seq is lost
+    // update last nacked
+    if (last_seq_nacked_ < seq) last_seq_nacked_ = seq;
+    TRANSPORT_LOGD("lost packet %u beacuse of a past nack", seq);
+    onPacketLost(seq);
+  } else if (seq > production_seq) {
+    // future nack
+    // remove the nack from the pending interest map
+    // (the packet is not received/lost yet)
+    pending_interests_.erase(seq);
+  } else {
+    // this should be a quite rear event. simply remove the
+    // packet from the pending interest list
+    pending_interests_.erase(seq);
+  }
+
+  // the producer is responding
+  // we consider it active only if the production rate is not 0
+  // or the production sequence number is not 1
+  if (production_rate_ != 0 || production_seq != 1) {
+    producer_is_active_ = true;
+  }
+
+  received_packets_last_round_++;
+}
+
+void RTCState::onPacketLost(uint32_t seq) {
+  TRANSPORT_LOGD("packet %u is lost", seq);
+  auto it = pending_interests_.find(seq);
+  if (it != pending_interests_.end()) {
+    // this packet was never retransmitted so it does
+    // not appear in the loss count
+    packets_lost_++;
+  }
+  addRecvOrLost(seq, PacketState::LOST);
+}
+
+void RTCState::onPacketRecovered(uint32_t seq) {
+  losses_recovered_++;
+  addRecvOrLost(seq, PacketState::RECEIVED);
+}
+
+bool RTCState::onProbePacketReceived(const core::ContentObject &probe) {
+  uint32_t seq = probe.getName().getSuffix();
+  uint64_t rtt;
+
+  rtt = rtt_probes_->getRtt(seq);
+
+  if (rtt == 0) return false;  // this is not a valid probe
+
+  // like for data and nacks update the path stats. Here the RTT is computed
+  // by the probe handler. Both probes for rtt and bw are good to esimate
+  // info on the path
+  uint32_t path_label = probe.getPathLabel();
+
+  auto path_it = path_table_.find(path_label);
+
+  // update production rate and last_seq_nacked like in case of a nack
+  struct nack_packet_t *probe_pkt =
+      (struct nack_packet_t *)probe.getPayload()->data();
+  uint64_t sender_timestamp = probe_pkt->getTimestamp();
+  uint32_t production_seq = probe_pkt->getProductionSegement();
+  uint32_t production_rate = probe_pkt->getProductionRate();
+
+
+  if (path_it == path_table_.end()) {
+    // found a new path
+    std::shared_ptr<RTCDataPath> newPath =
+        std::make_shared<RTCDataPath>(path_label);
+    auto ret = path_table_.insert(
+        std::pair<uint32_t, std::shared_ptr<RTCDataPath>>(path_label, newPath));
+    path_it = ret.first;
+  }
+
+  auto path = path_it->second;
+
+  path->insertRttSample(rtt);
+  path->receivedNack();
+
+  uint64_t now = std::chrono::duration_cast<std::chrono::milliseconds>(
+                     std::chrono::steady_clock::now().time_since_epoch())
+                     .count();
+
+  int64_t OWD = now - sender_timestamp;
+  path->insertOwdSample(OWD);
+
+  if (last_prod_update_ < sender_timestamp) {
+    last_production_seq_ = production_seq;
+    last_prod_update_ = sender_timestamp;
+    production_rate_ = (double)production_rate;
+  }
+
+  // the producer is responding
+  // we consider it active only if the production rate is not 0
+  // or the production sequence numner is not 1
+  if (production_rate_ != 0 || production_seq != 1) {
+    producer_is_active_ = true;
+  }
+
+  // check for init RTT. if received_probes_ is equal to 0 schedule a timer to
+  // wait for the INIT_RTT_PROBES. in this way if some probes get lost we don't
+  // wait forever
+  received_probes_++;
+
+  if(!init_rtt_ && received_probes_ <= INIT_RTT_PROBES){
+    if(received_probes_ == 1){
+      // we got the first probe, wait at most INIT_RTT_PROBE_WAIT sec for the others
+      main_path_ = path;
+      setInitRttTimer(INIT_RTT_PROBE_WAIT);
+    }
+    if(received_probes_ == INIT_RTT_PROBES) {
+      // we are done
+      init_rtt_timer_->cancel();
+      checkInitRttTimer();
+    }
+  }
+
+  received_packets_last_round_++;
+
+  // ignore probes sent before the first interest
+  if((now - rtt) <= first_interest_sent_) return false;
+  return true;
+}
+
+void RTCState::onNewRound(double round_len, bool in_sync) {
+  // XXX
+  // here we take into account only the single path case so we assume that we
+  // don't use two paths in parellel for this single flow
+
+  if (path_table_.empty()) return;
+
+  double bytes_per_sec =
+      ((double)received_bytes_ * (MILLI_IN_A_SEC / round_len));
+  if(received_rate_ == 0)
+    received_rate_ = bytes_per_sec;
+  else
+    received_rate_ = (received_rate_ * MOVING_AVG_ALPHA) +
+                   ((1 - MOVING_AVG_ALPHA) * bytes_per_sec);
+
+  // search for an active path. There should be only one active path (meaning a
+  // path that leads to the producer socket -no cache- and from which we are
+  // currently getting data packets) at any time. However it may happen that
+  // there are mulitple active paths in case of mobility (the old path will
+  // remain active for a short ammount of time). The main path is selected as
+  // the active path from where the consumer received the latest data packet
+
+  uint64_t last_packet_ts = 0;
+  main_path_ = nullptr;
+
+  for (auto it = path_table_.begin(); it != path_table_.end(); it++) {
+    it->second->roundEnd();
+    if (it->second->isActive()) {
+      uint64_t ts = it->second->getLastPacketTS();
+      if (ts > last_packet_ts) {
+        last_packet_ts = ts;
+        main_path_ = it->second;
+      }
+    }
+  }
+
+  if (in_sync) updateLossRate();
+
+  // handle nacks
+  if (!nack_on_last_round_ && received_bytes_ > 0) {
+    rounds_without_nacks_++;
+  } else {
+    rounds_without_nacks_ = 0;
+  }
+
+  // check if the producer is active
+  if (received_packets_last_round_ != 0) {
+    rounds_without_packets_ = 0;
+  } else {
+    rounds_without_packets_++;
+    if (rounds_without_packets_ >= MAX_ROUND_WHIOUT_PACKETS &&
+        producer_is_active_ != false) {
+      initParams();
+    }
+  }
+
+  // compute cache/producer ratio
+  if (received_data_last_round_ != 0) {
+    double new_rate =
+        (double)received_data_from_cache_ / (double)received_data_last_round_;
+    data_from_cache_rate_ = data_from_cache_rate_ * MOVING_AVG_ALPHA +
+                            (new_rate * (1 - MOVING_AVG_ALPHA));
+  }
+
+  // reset counters
+  received_bytes_ = 0;
+  packets_lost_ = 0;
+  losses_recovered_ = 0;
+  first_seq_in_round_ = highest_seq_received_;
+
+  nack_on_last_round_ = false;
+  received_nacks_last_round_ = 0;
+
+  received_packets_last_round_ = 0;
+  received_data_last_round_ = 0;
+  received_data_from_cache_ = 0;
+  sent_interests_last_round_ = 0;
+  sent_rtx_last_round_ = 0;
+
+  rounds_++;
+}
+
+void RTCState::updateReceivedBytes(const core::ContentObject &content_object) {
+  received_bytes_ +=
+      (uint32_t)(content_object.headerSize() + content_object.payloadSize());
+}
+
+void RTCState::updatePacketSize(const core::ContentObject &content_object) {
+  uint32_t pkt_size =
+      (uint32_t)(content_object.headerSize() + content_object.payloadSize());
+  avg_packet_size_ = (MOVING_AVG_ALPHA * avg_packet_size_) +
+                     ((1 - MOVING_AVG_ALPHA) * pkt_size);
+}
+
+void RTCState::updatePathStats(const core::ContentObject &content_object,
+                               bool is_nack) {
+  // get packet path
+  uint32_t path_label = content_object.getPathLabel();
+  auto path_it = path_table_.find(path_label);
+
+  if (path_it == path_table_.end()) {
+    // found a new path
+    std::shared_ptr<RTCDataPath> newPath =
+        std::make_shared<RTCDataPath>(path_label);
+    auto ret = path_table_.insert(
+        std::pair<uint32_t, std::shared_ptr<RTCDataPath>>(path_label, newPath));
+    path_it = ret.first;
+  }
+
+  auto path = path_it->second;
+
+  // compute rtt
+  uint32_t seq = content_object.getName().getSuffix();
+  uint64_t interest_sent_time = getInterestSentTime(seq);
+  if (interest_sent_time == 0)
+    return;  // this should not happen,
+             // it means that we are processing an interest
+             // that is not pending
+
+  uint64_t now = std::chrono::duration_cast<std::chrono::milliseconds>(
+                     std::chrono::steady_clock::now().time_since_epoch())
+                     .count();
+
+  uint64_t RTT = now - interest_sent_time;
+
+  path->insertRttSample(RTT);
+
+  // compute OWD (the first part of the nack and data packet header are the
+  // same, so we cast to data data packet)
+  struct data_packet_t *packet =
+      (struct data_packet_t *)content_object.getPayload()->data();
+  uint64_t sender_timestamp = packet->getTimestamp();
+  int64_t OWD = now - sender_timestamp;
+  path->insertOwdSample(OWD);
+
+  // compute IAT or set path to producer
+  if (!is_nack) {
+    // compute the iat only for the content packets
+    uint32_t segment_number = content_object.getName().getSuffix();
+    path->computeInterArrivalGap(segment_number);
+    if (!path->pathToProducer()) received_data_from_cache_++;
+  } else {
+    path->receivedNack();
+  }
+}
+
+void RTCState::updateLossRate() {
+  loss_rate_ = 0.0;
+  residual_loss_rate_ = 0.0;
+
+  uint32_t number_theorically_received_packets_ =
+      highest_seq_received_ - first_seq_in_round_;
+
+  // in this case no new packet was recevied after the previuos round, avoid
+  // division by 0
+  if (number_theorically_received_packets_ == 0) return;
+
+  loss_rate_ = (double)((double)(packets_lost_) /
+                        (double)number_theorically_received_packets_);
+
+  residual_loss_rate_ = (double)((double)(packets_lost_ - losses_recovered_) /
+                                 (double)number_theorically_received_packets_);
+
+  if (residual_loss_rate_ < 0) residual_loss_rate_ = 0;
+}
+
+void RTCState::addRecvOrLost(uint32_t seq, PacketState state) {
+  pending_interests_.erase(seq);
+  if (received_or_lost_packets_.size() >= MAX_CACHED_PACKETS) {
+    received_or_lost_packets_.erase(received_or_lost_packets_.begin());
+  }
+  // notice that it may happen that a packet that we consider lost arrives after
+  // some time, in this case we simply overwrite the packet state.
+  received_or_lost_packets_[seq] = state;
+
+  // keep track of the last packet received/lost
+  // without holes.
+  if (highest_seq_received_in_order_ < last_seq_nacked_) {
+    highest_seq_received_in_order_ = last_seq_nacked_;
+  }
+
+  if ((highest_seq_received_in_order_ + 1) == seq) {
+    highest_seq_received_in_order_ = seq;
+  } else if (seq <= highest_seq_received_in_order_) {
+    // here we do nothing
+  } else if (seq > highest_seq_received_in_order_) {
+    // 1) there is a gap in the sequence so we do not update largest_in_seq_
+    // 2) all the packets from largest_in_seq_ to seq are in
+    //    received_or_lost_packets_ an we upate largest_in_seq_
+
+    for (uint32_t i = highest_seq_received_in_order_ + 1; i <= seq; i++) {
+      if (received_or_lost_packets_.find(i) ==
+          received_or_lost_packets_.end()) {
+        break;
+      }
+      // this packet is in order so we can update the
+      // highest_seq_received_in_order_
+      highest_seq_received_in_order_ = i;
+    }
+  }
+}
+
+void RTCState::setInitRttTimer(uint32_t wait){
+  init_rtt_timer_->cancel();
+  init_rtt_timer_->expires_from_now(std::chrono::milliseconds(wait));
+  init_rtt_timer_->async_wait([this](std::error_code ec) {
+    if(ec) return;
+    checkInitRttTimer();
+  });
+}
+
+void RTCState::checkInitRttTimer() {
+  if(received_probes_ < INIT_RTT_MIN_PROBES_TO_RECV){
+    // we didn't received enough probes, restart
+    received_probes_ = 0;
+    rtt_probes_->setProbes(INIT_RTT_PROBE_INTERVAL, INIT_RTT_PROBES);
+    rtt_probes_->sendProbes();
+    setInitRttTimer(INIT_RTT_PROBE_RESTART);
+    return;
+  }
+  init_rtt_ = true;
+  main_path_->roundEnd();
+  rtt_probes_->setProbes(RTT_PROBE_INTERVAL, 0);
+  rtt_probes_->sendProbes();
+
+  // init last_seq_nacked_. skip packets that may come from the cache
+  double prod_rate = getProducerRate();
+  double rtt = (double)getRTT() / MILLI_IN_A_SEC;
+  double packet_size = getAveragePacketSize();
+  uint32_t pkt_in_rtt_ = std::floor(((prod_rate / packet_size) * rtt) * 0.8);
+  last_seq_nacked_ = last_production_seq_ + pkt_in_rtt_;
+
+  discovered_rtt_callback_();
+}
+
+}  // namespace rtc
+
+}  // namespace protocol
+
+}  // namespace transport