tcp: increase flexibility of EBPF congestion control initialization

.travis.yml

@@ -0,0 +1,32 @@
+sudo: required
+language: bash
+dist: bionic
+services:
+    - docker
+
+env:
+    global:
+        - PROJECT_NAME='libbpf'
+        - AUTHOR_EMAIL="$(git log -1 --pretty=\"%aE\")"
+        - REPO_ROOT="$TRAVIS_BUILD_DIR"
+        - CI_ROOT="$REPO_ROOT/travis-ci"
+        - VMTEST_ROOT="$CI_ROOT/vmtest"
+
+addons:
+    apt:
+        packages:
+            - qemu-kvm
+            - zstd
+            - binutils-dev
+            - elfutils
+            - libcap-dev
+            - libelf-dev
+            - libdw-dev
+
+jobs:
+    include:
+        - stage: Builds & Tests
+          name: Kernel LATEST + selftests
+          language: bash
+          env: KERNEL=LATEST
+          script:  $CI_ROOT/vmtest/run_vmtest.sh || travis_terminate 1

include/net/inet_connection_sock.h

@@ -96,7 +96,8 @@ struct inet_connection_sock {
 	void (*icsk_clean_acked)(struct sock *sk, u32 acked_seq);
 	struct hlist_node         icsk_listen_portaddr_node;
 	unsigned int		  (*icsk_sync_mss)(struct sock *sk, u32 pmtu);
-	__u8			  icsk_ca_state:6,
+	__u8			  icsk_ca_state:5,
+				  icsk_ca_initialized:1,
 				  icsk_ca_setsockopt:1,
 				  icsk_ca_dst_locked:1;
 	__u8			  icsk_retransmits;

include/net/tcp.h

@@ -1104,7 +1104,7 @@ void tcp_get_available_congestion_control(char *buf, size_t len);
 void tcp_get_allowed_congestion_control(char *buf, size_t len);
 int tcp_set_allowed_congestion_control(char *allowed);
 int tcp_set_congestion_control(struct sock *sk, const char *name, bool load,
-			       bool reinit, bool cap_net_admin);
+			       bool cap_net_admin);
 u32 tcp_slow_start(struct tcp_sock *tp, u32 acked);
 void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked);
 

net/core/filter.c

@@ -4313,10 +4313,8 @@ static const struct bpf_func_proto bpf_get_socket_uid_proto = {
 	.arg1_type      = ARG_PTR_TO_CTX,
 };
 
-#define SOCKOPT_CC_REINIT (1 << 0)
-
 static int _bpf_setsockopt(struct sock *sk, int level, int optname,
-			   char *optval, int optlen, u32 flags)
+			   char *optval, int optlen)
 {
 	char devname[IFNAMSIZ];
 	int val, valbool;
@@ -4449,13 +4447,11 @@ static int _bpf_setsockopt(struct sock *sk, int level, int optname,
 		   sk->sk_prot->setsockopt == tcp_setsockopt) {
 		if (optname == TCP_CONGESTION) {
 			char name[TCP_CA_NAME_MAX];
-			bool reinit = flags & SOCKOPT_CC_REINIT;
 
 			strncpy(name, optval, min_t(long, optlen,
 						    TCP_CA_NAME_MAX-1));
 			name[TCP_CA_NAME_MAX-1] = 0;
-			ret = tcp_set_congestion_control(sk, name, false,
-							 reinit, true);
+			ret = tcp_set_congestion_control(sk, name, false, true);
 		} else {
 			struct inet_connection_sock *icsk = inet_csk(sk);
 			struct tcp_sock *tp = tcp_sk(sk);
@@ -4615,9 +4611,7 @@ static int _bpf_getsockopt(struct sock *sk, int level, int optname,
 BPF_CALL_5(bpf_sock_addr_setsockopt, struct bpf_sock_addr_kern *, ctx,
 	   int, level, int, optname, char *, optval, int, optlen)
 {
-	u32 flags = 0;
-	return _bpf_setsockopt(ctx->sk, level, optname, optval, optlen,
-			       flags);
+	return _bpf_setsockopt(ctx->sk, level, optname, optval, optlen);
 }
 
 static const struct bpf_func_proto bpf_sock_addr_setsockopt_proto = {
@@ -4651,11 +4645,7 @@ static const struct bpf_func_proto bpf_sock_addr_getsockopt_proto = {
 BPF_CALL_5(bpf_sock_ops_setsockopt, struct bpf_sock_ops_kern *, bpf_sock,
 	   int, level, int, optname, char *, optval, int, optlen)
 {
-	u32 flags = 0;
-	if (bpf_sock->op > BPF_SOCK_OPS_NEEDS_ECN)
-		flags |= SOCKOPT_CC_REINIT;
-	return _bpf_setsockopt(bpf_sock->sk, level, optname, optval, optlen,
-			       flags);
+	return _bpf_setsockopt(bpf_sock->sk, level, optname, optval, optlen);
 }
 
 static const struct bpf_func_proto bpf_sock_ops_setsockopt_proto = {

net/ipv4/tcp.c

@@ -2698,6 +2698,7 @@ int tcp_disconnect(struct sock *sk, int flags)
 	if (icsk->icsk_ca_ops->release)
 		icsk->icsk_ca_ops->release(sk);
 	memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
+	icsk->icsk_ca_initialized = 0;
 	tcp_set_ca_state(sk, TCP_CA_Open);
 	tp->is_sack_reneg = 0;
 	tcp_clear_retrans(tp);
@@ -3049,7 +3050,7 @@ static int do_tcp_setsockopt(struct sock *sk, int level, int optname,
 		name[val] = 0;
 
 		lock_sock(sk);
-		err = tcp_set_congestion_control(sk, name, true, true,
+		err = tcp_set_congestion_control(sk, name, true,
 						 ns_capable(sock_net(sk)->user_ns,
 							    CAP_NET_ADMIN));
 		release_sock(sk);

net/ipv4/tcp_cong.c

@@ -176,7 +176,7 @@ void tcp_assign_congestion_control(struct sock *sk)
 
 void tcp_init_congestion_control(struct sock *sk)
 {
-	const struct inet_connection_sock *icsk = inet_csk(sk);
+	struct inet_connection_sock *icsk = inet_csk(sk);
 
 	tcp_sk(sk)->prior_ssthresh = 0;
 	if (icsk->icsk_ca_ops->init)
@@ -185,6 +185,7 @@ void tcp_init_congestion_control(struct sock *sk)
 		INET_ECN_xmit(sk);
 	else
 		INET_ECN_dontxmit(sk);
+	icsk->icsk_ca_initialized = 1;
 }
 
 static void tcp_reinit_congestion_control(struct sock *sk,
@@ -340,7 +341,7 @@ int tcp_set_allowed_congestion_control(char *val)
  * already initialized.
  */
 int tcp_set_congestion_control(struct sock *sk, const char *name, bool load,
-			       bool reinit, bool cap_net_admin)
+			       bool cap_net_admin)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 	const struct tcp_congestion_ops *ca;
@@ -361,28 +362,14 @@ int tcp_set_congestion_control(struct sock *sk, const char *name, bool load,
 		goto out;
 	}
 
-	if (!ca) {
+	if (!ca)
 		err = -ENOENT;
-	} else if (!load) {
-		const struct tcp_congestion_ops *old_ca = icsk->icsk_ca_ops;
-
-		if (bpf_try_module_get(ca, ca->owner)) {
-			if (reinit) {
-				tcp_reinit_congestion_control(sk, ca);
-			} else {
-				icsk->icsk_ca_ops = ca;
-				bpf_module_put(old_ca, old_ca->owner);
-			}
-		} else {
-			err = -EBUSY;
-		}
-	} else if (!((ca->flags & TCP_CONG_NON_RESTRICTED) || cap_net_admin)) {
+	else if (!((ca->flags & TCP_CONG_NON_RESTRICTED) || cap_net_admin))
 		err = -EPERM;
-	} else if (!bpf_try_module_get(ca, ca->owner)) {
+	else if (!bpf_try_module_get(ca, ca->owner))
 		err = -EBUSY;
-	} else {
+	else
 		tcp_reinit_congestion_control(sk, ca);
-	}
  out:
 	rcu_read_unlock();
 	return err;

net/ipv4/tcp_input.c

@@ -5894,8 +5894,10 @@ void tcp_init_transfer(struct sock *sk, int bpf_op, struct sk_buff *skb)
 		tp->snd_cwnd = tcp_init_cwnd(tp, __sk_dst_get(sk));
 	tp->snd_cwnd_stamp = tcp_jiffies32;
 
+	icsk->icsk_ca_initialized = 0;
 	bpf_skops_established(sk, bpf_op, skb);
-	tcp_init_congestion_control(sk);
+	if (!icsk->icsk_ca_initialized)
+		tcp_init_congestion_control(sk);
 	tcp_init_buffer_space(sk);
 }
 

travis-ci/managers/debian.sh

@@ -0,0 +1,84 @@
+#!/bin/bash
+
+PHASES=(${@:-SETUP RUN RUN_ASAN CLEANUP})
+DEBIAN_RELEASE="${DEBIAN_RELEASE:-testing}"
+CONT_NAME="${CONT_NAME:-libbpf-debian-$DEBIAN_RELEASE}"
+ENV_VARS="${ENV_VARS:-}"
+DOCKER_RUN="${DOCKER_RUN:-docker run}"
+REPO_ROOT="${REPO_ROOT:-$PWD}"
+ADDITIONAL_DEPS=(clang pkg-config gcc-8)
+CFLAGS="-g -O2 -Werror -Wall"
+
+function info() {
+    echo -e "\033[33;1m$1\033[0m"
+}
+
+function error() {
+    echo -e "\033[31;1m$1\033[0m"
+}
+
+function docker_exec() {
+    docker exec $ENV_VARS -it $CONT_NAME "$@"
+}
+
+set -e
+
+source "$(dirname $0)/travis_wait.bash"
+
+for phase in "${PHASES[@]}"; do
+    case $phase in
+        SETUP)
+            info "Setup phase"
+            info "Using Debian $DEBIAN_RELEASE"
+
+            sudo apt-get -y -o Dpkg::Options::="--force-confnew" install docker-ce
+            docker --version
+
+            docker pull debian:$DEBIAN_RELEASE
+            info "Starting container $CONT_NAME"
+            $DOCKER_RUN -v $REPO_ROOT:/build:rw \
+                        -w /build --privileged=true --name $CONT_NAME \
+                        -dit --net=host debian:$DEBIAN_RELEASE /bin/bash
+            docker_exec bash -c "echo deb-src http://deb.debian.org/debian $DEBIAN_RELEASE main >>/etc/apt/sources.list"
+            docker_exec apt-get -y update
+            docker_exec apt-get -y build-dep libelf-dev
+            docker_exec apt-get -y install libelf-dev
+            docker_exec apt-get -y install "${ADDITIONAL_DEPS[@]}"
+            ;;
+        RUN|RUN_CLANG|RUN_GCC8|RUN_ASAN|RUN_CLANG_ASAN|RUN_GCC8_ASAN)
+            if [[ "$phase" = *"CLANG"* ]]; then
+                ENV_VARS="-e CC=clang -e CXX=clang++"
+                CC="clang"
+            elif [[ "$phase" = *"GCC8"* ]]; then
+                ENV_VARS="-e CC=gcc-8 -e CXX=g++-8"
+                CC="gcc-8"
+            else
+                CFLAGS="${CFLAGS} -Wno-stringop-truncation"
+            fi
+            if [[ "$phase" = *"ASAN"* ]]; then
+                CFLAGS="${CFLAGS} -fsanitize=address,undefined"
+            fi
+            docker_exec mkdir build install
+            docker_exec ${CC:-cc} --version
+            info "build"
+	    docker_exec make -j$((4*$(nproc))) CFLAGS="${CFLAGS}" -C ./src -B OBJDIR=../build
+            info "ldd build/libbpf.so:"
+            docker_exec ldd build/libbpf.so
+            if ! docker_exec ldd build/libbpf.so | grep -q libelf; then
+                error "No reference to libelf.so in libbpf.so!"
+                exit 1
+            fi
+            info "install"
+            docker_exec make -j$((4*$(nproc))) -C src OBJDIR=../build DESTDIR=../install install
+            docker_exec rm -rf build install
+            ;;
+        CLEANUP)
+            info "Cleanup phase"
+            docker stop $CONT_NAME
+            docker rm -f $CONT_NAME
+            ;;
+        *)
+            echo >&2 "Unknown phase '$phase'"
+            exit 1
+    esac
+done

travis-ci/managers/travis_wait.bash

@@ -0,0 +1,61 @@
+# This was borrowed from https://github.com/travis-ci/travis-build/tree/master/lib/travis/build/bash
+# to get around https://github.com/travis-ci/travis-ci/issues/9979. It should probably be removed
+# as soon as Travis CI has started to provide an easy way to export the functions to bash scripts.
+
+travis_jigger() {
+  local cmd_pid="${1}"
+  shift
+  local timeout="${1}"
+  shift
+  local count=0
+
+  echo -e "\\n"
+
+  while [[ "${count}" -lt "${timeout}" ]]; do
+    count="$((count + 1))"
+    echo -ne "Still running (${count} of ${timeout}): ${*}\\r"
+    sleep 60
+  done
+
+  echo -e "\\n${ANSI_RED}Timeout (${timeout} minutes) reached. Terminating \"${*}\"${ANSI_RESET}\\n"
+  kill -9 "${cmd_pid}"
+}
+
+travis_wait() {
+  local timeout="${1}"
+
+  if [[ "${timeout}" =~ ^[0-9]+$ ]]; then
+    shift
+  else
+    timeout=20
+  fi
+
+  local cmd=("${@}")
+  local log_file="travis_wait_${$}.log"
+
+  "${cmd[@]}" &>"${log_file}" &
+  local cmd_pid="${!}"
+
+  travis_jigger "${!}" "${timeout}" "${cmd[@]}" &
+  local jigger_pid="${!}"
+  local result
+
+  {
+    set +e
+    wait "${cmd_pid}" 2>/dev/null
+    result="${?}"
+    ps -p"${jigger_pid}" &>/dev/null && kill "${jigger_pid}"
+    set -e
+  }
+
+  if [[ "${result}" -eq 0 ]]; then
+    echo -e "\\n${ANSI_GREEN}The command ${cmd[*]} exited with ${result}.${ANSI_RESET}"
+  else
+    echo -e "\\n${ANSI_RED}The command ${cmd[*]} exited with ${result}.${ANSI_RESET}"
+  fi
+
+  echo -e "\\n${ANSI_GREEN}Log:${ANSI_RESET}\\n"
+  cat "${log_file}"
+
+  return "${result}"
+}

travis-ci/managers/ubuntu.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+set -e
+set -x
+
+RELEASE="bionic"
+
+echo "deb-src http://archive.ubuntu.com/ubuntu/ $RELEASE main restricted universe multiverse" >>/etc/apt/sources.list
+
+apt-get update
+apt-get -y build-dep libelf-dev
+apt-get install -y libelf-dev pkg-config
+
+source "$(dirname $0)/travis_wait.bash"
+
+cd $REPO_ROOT
+
+CFLAGS="-g -O2 -Werror -Wall -fsanitize=address,undefined"
+mkdir build install
+cc --version
+make -j$((4*$(nproc))) CFLAGS="${CFLAGS}" -C ./src -B OBJDIR=../build
+ldd build/libbpf.so
+if ! ldd build/libbpf.so | grep -q libelf; then
+    echo "FAIL: No reference to libelf.so in libbpf.so!"
+    exit 1
+fi
+make -j$((4*$(nproc))) -C src OBJDIR=../build DESTDIR=../install install
+rm -rf build install

travis-ci/vmtest/build_pahole.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+
+set -eu
+
+source $(cd $(dirname $0) && pwd)/helpers.sh
+
+travis_fold start build_pahole "Building pahole"
+
+CWD=$(pwd)
+REPO_PATH=$1
+PAHOLE_ORIGIN=https://git.kernel.org/pub/scm/devel/pahole/pahole.git
+
+mkdir -p ${REPO_PATH}
+cd ${REPO_PATH}
+git init
+git remote add origin ${PAHOLE_ORIGIN}
+git fetch origin
+git checkout master
+
+mkdir -p build
+cd build
+cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo -D__LIB=lib ..
+make -j$((4*$(nproc))) all
+sudo make install
+
+export LD_LIBRARY_PATH=${LD_LIBRARY_PATH:-}:/usr/local/lib
+ldd $(which pahole)
+pahole --version
+
+travis_fold end build_pahole