As reported by Mat, the in kernel PM can, in some edge scenarios, unexpectedly create multiple subflows with the same local and remote address. The real fix is implemented by patch 4/4 with some more accurate check at subflow creation time. Patches 1-3 are roughly optional pre-requisities, added to avoid introducing more data-races with the actual fix. Patch 1/4 is a bit debatable, as it changes the existing ULP API, but I could not find a better solution and there is some similar prior art: commit 0df48c26d841 ("tcp: add tcpi_bytes_acked to tcp_info") Paolo Abeni (4): mptcp: fix lockless access in subflow ULP diag mptcp: fix data races on local_id mptcp: fix data races on remote_id mptcp: fix duplicate subflow creation include/net/tcp.h | 2 +- net/mptcp/diag.c | 6 +++++- net/mptcp/pm_netlink.c | 43 ++++++++++++++++++++++-------------------- net/mptcp/protocol.c | 2 +- net/mptcp/protocol.h | 15 ++++++++++++--- net/mptcp/subflow.c | 15 ++++++++------- net/tls/tls_main.c | 2 +- 7 files changed, 51 insertions(+), 34 deletions(-) -- 2.43.0 Since the introduction of the subflow ULP diag interface, the dump callback accessed all the subflow data with lockless. We need either to annotate all the read and write operation accordingly, or acquire the subflow socket lock. Let's do latter, even if slower, to avoid a diffstat havoc. Fixes: 5147dfb50832 ("mptcp: allow dumping subflow context to userspace") Signed-off-by: Paolo Abeni <pabeni@redhat.com> --- note: tls ulp diag has likely the same issue. --- include/net/tcp.h | 2 +- net/mptcp/diag.c | 6 +++++- net/tls/tls_main.c | 2 +- 3 files changed, 7 insertions(+), 3 deletions(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index XXXXXXX..XXXXXXX 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -XXX,XX +XXX,XX @@ struct tcp_ulp_ops { /* cleanup ulp */ void (*release)(struct sock *sk); /* diagnostic */ - int (*get_info)(const struct sock *sk, struct sk_buff *skb); + int (*get_info)(struct sock *sk, struct sk_buff *skb); size_t (*get_info_size)(const struct sock *sk); /* clone ulp */ void (*clone)(const struct request_sock *req, struct sock *newsk, diff --git a/net/mptcp/diag.c b/net/mptcp/diag.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/diag.c +++ b/net/mptcp/diag.c @@ -XXX,XX +XXX,XX @@ #include <uapi/linux/mptcp.h> #include "protocol.h" -static int subflow_get_info(const struct sock *sk, struct sk_buff *skb) +static int subflow_get_info(struct sock *sk, struct sk_buff *skb) { struct mptcp_subflow_context *sf; struct nlattr *start; u32 flags = 0; + bool slow; int err; start = nla_nest_start_noflag(skb, INET_ULP_INFO_MPTCP); if (!start) return -EMSGSIZE; + slow = lock_sock_fast(sk); rcu_read_lock(); sf = rcu_dereference(inet_csk(sk)->icsk_ulp_data); if (!sf) { @@ -XXX,XX +XXX,XX @@ static int subflow_get_info(const struct sock *sk, struct sk_buff *skb) } rcu_read_unlock(); + unlock_sock_fast(sk, slow); nla_nest_end(skb, start); return 0; nla_failure: rcu_read_unlock(); + unlock_sock_fast(sk, slow); nla_nest_cancel(skb, start); return err; } diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c index XXXXXXX..XXXXXXX 100644 --- a/net/tls/tls_main.c +++ b/net/tls/tls_main.c @@ -XXX,XX +XXX,XX @@ static u16 tls_user_config(struct tls_context *ctx, bool tx) return 0; } -static int tls_get_info(const struct sock *sk, struct sk_buff *skb) +static int tls_get_info(struct sock *sk, struct sk_buff *skb) { u16 version, cipher_type; struct tls_context *ctx; -- 2.43.0 The local address id is accessed lockless by the NL PM, add all the required ONCE annotation. There is a caveat: the local id can be initialized late in the subflow life-cycle, and its validity is controlled by the local_id_valid flag. Remove such flag and encode the validity in the local_id field itself with negative value before initialization. That allows accessing the field consistently with a single read operation. Fixes: 0ee4261a3681 ("mptcp: implement mptcp_pm_remove_subflow") Signed-off-by: Paolo Abeni <pabeni@redhat.com> --- net/mptcp/pm_netlink.c | 4 ++-- net/mptcp/protocol.c | 2 +- net/mptcp/protocol.h | 15 ++++++++++++--- net/mptcp/subflow.c | 9 +++++---- 4 files changed, 20 insertions(+), 10 deletions(-) diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -XXX,XX +XXX,XX @@ static void mptcp_pm_nl_rm_addr_or_subflow(struct mptcp_sock *msk, mptcp_for_each_subflow_safe(msk, subflow, tmp) { struct sock *ssk = mptcp_subflow_tcp_sock(subflow); int how = RCV_SHUTDOWN | SEND_SHUTDOWN; - u8 id = subflow->local_id; + u8 id = subflow_get_local_id(subflow); if (rm_type == MPTCP_MIB_RMADDR && subflow->remote_id != rm_id) continue; @@ -XXX,XX +XXX,XX @@ static void mptcp_pm_nl_rm_addr_or_subflow(struct mptcp_sock *msk, pr_debug(" -> %s rm_list_ids[%d]=%u local_id=%u remote_id=%u mpc_id=%u", rm_type == MPTCP_MIB_RMADDR ? "address" : "subflow", - i, rm_id, subflow->local_id, subflow->remote_id, + i, rm_id, id, subflow->remote_id, msk->mpc_endpoint_id); spin_unlock_bh(&msk->pm.lock); mptcp_subflow_shutdown(sk, ssk, how); diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -XXX,XX +XXX,XX @@ static int __mptcp_socket_create(struct mptcp_sock *msk) subflow->subflow_id = msk->subflow_id++; /* This is the first subflow, always with id 0 */ - subflow->local_id_valid = 1; + WRITE_ONCE(subflow->local_id, 0); mptcp_sock_graft(msk->first, sk->sk_socket); iput(SOCK_INODE(ssock)); diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -XXX,XX +XXX,XX @@ struct mptcp_subflow_context { remote_key_valid : 1, /* received the peer key from */ disposable : 1, /* ctx can be free at ulp release time */ stale : 1, /* unable to snd/rcv data, do not use for xmit */ - local_id_valid : 1, /* local_id is correctly initialized */ valid_csum_seen : 1, /* at least one csum validated */ is_mptfo : 1, /* subflow is doing TFO */ - __unused : 9; + __unused : 10; bool data_avail; bool scheduled; u32 remote_nonce; @@ -XXX,XX +XXX,XX @@ struct mptcp_subflow_context { u8 hmac[MPTCPOPT_HMAC_LEN]; /* MPJ subflow only */ u64 iasn; /* initial ack sequence number, MPC subflows only */ }; - u8 local_id; + s16 local_id; /* if negative not initialized yet */ u8 remote_id; u8 reset_seen:1; u8 reset_transient:1; @@ -XXX,XX +XXX,XX @@ mptcp_subflow_ctx_reset(struct mptcp_subflow_context *subflow) { memset(&subflow->reset, 0, sizeof(subflow->reset)); subflow->request_mptcp = 1; + WRITE_ONCE(subflow->local_id, -1); } static inline u64 @@ -XXX,XX +XXX,XX @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc); int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc); int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc); +static inline int subflow_get_local_id(struct mptcp_subflow_context *subflow) +{ + int local_id = READ_ONCE(subflow->local_id); + + if (local_id < 0) + return 0; + return local_id; +} + void __init mptcp_pm_nl_init(void); void mptcp_pm_nl_work(struct mptcp_sock *msk); void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk, diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -XXX,XX +XXX,XX @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb) static void subflow_set_local_id(struct mptcp_subflow_context *subflow, int local_id) { - subflow->local_id = local_id; - subflow->local_id_valid = 1; + WARN_ON_ONCE(local_id < 0 || local_id > 255); + WRITE_ONCE(subflow->local_id, local_id); } static int subflow_chk_local_id(struct sock *sk) @@ -XXX,XX +XXX,XX @@ static int subflow_chk_local_id(struct sock *sk) struct mptcp_sock *msk = mptcp_sk(subflow->conn); int err; - if (likely(subflow->local_id_valid)) + if (likely(subflow->local_id >= 0)) return 0; err = mptcp_pm_get_local_id(msk, (struct sock_common *)sk); @@ -XXX,XX +XXX,XX @@ static struct mptcp_subflow_context *subflow_create_ctx(struct sock *sk, pr_debug("subflow=%p", ctx); ctx->tcp_sock = sk; + WRITE_ONCE(ctx->local_id, -1); return ctx; } @@ -XXX,XX +XXX,XX @@ static void subflow_ulp_clone(const struct request_sock *req, new_ctx->idsn = subflow_req->idsn; /* this is the first subflow, id is always 0 */ - new_ctx->local_id_valid = 1; + subflow_set_local_id(new_ctx, 0); } else if (subflow_req->mp_join) { new_ctx->ssn_offset = subflow_req->ssn_offset; new_ctx->mp_join = 1; -- 2.43.0 Similar to the previous patch, address the data race on remote_id, adding the suitable ONCE annotations. Fixes: bedee0b56113 ("mptcp: address lookup improvements") Signed-off-by: Paolo Abeni <pabeni@redhat.com> --- net/mptcp/pm_netlink.c | 8 ++++---- net/mptcp/subflow.c | 6 +++--- 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -XXX,XX +XXX,XX @@ static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, mptcp_for_each_subflow(msk, subflow) { ssk = mptcp_subflow_tcp_sock(subflow); remote_address((struct sock_common *)ssk, &addrs[i]); - addrs[i].id = subflow->remote_id; + addrs[i].id = READ_ONCE(subflow->remote_id); if (deny_id0 && !addrs[i].id) continue; @@ -XXX,XX +XXX,XX @@ static void mptcp_pm_nl_rm_addr_or_subflow(struct mptcp_sock *msk, mptcp_for_each_subflow_safe(msk, subflow, tmp) { struct sock *ssk = mptcp_subflow_tcp_sock(subflow); int how = RCV_SHUTDOWN | SEND_SHUTDOWN; + u8 srm_id = READ_ONCE(subflow->remote_id); u8 id = subflow_get_local_id(subflow); - if (rm_type == MPTCP_MIB_RMADDR && subflow->remote_id != rm_id) + if (rm_type == MPTCP_MIB_RMADDR && srm_id != rm_id) continue; if (rm_type == MPTCP_MIB_RMSUBFLOW && !mptcp_local_id_match(msk, id, rm_id)) continue; pr_debug(" -> %s rm_list_ids[%d]=%u local_id=%u remote_id=%u mpc_id=%u", rm_type == MPTCP_MIB_RMADDR ? "address" : "subflow", - i, rm_id, id, subflow->remote_id, - msk->mpc_endpoint_id); + i, rm_id, id, srm_id, msk->mpc_endpoint_id); spin_unlock_bh(&msk->pm.lock); mptcp_subflow_shutdown(sk, ssk, how); diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -XXX,XX +XXX,XX @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb) subflow->backup = mp_opt.backup; subflow->thmac = mp_opt.thmac; subflow->remote_nonce = mp_opt.nonce; - subflow->remote_id = mp_opt.join_id; + WRITE_ONCE(subflow->remote_id, mp_opt.join_id); pr_debug("subflow=%p, thmac=%llu, remote_nonce=%u backup=%d", subflow, subflow->thmac, subflow->remote_nonce, subflow->backup); @@ -XXX,XX +XXX,XX @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc, pr_debug("msk=%p remote_token=%u local_id=%d remote_id=%d", msk, remote_token, local_id, remote_id); subflow->remote_token = remote_token; - subflow->remote_id = remote_id; + WRITE_ONCE(subflow->remote_id, remote_id); subflow->request_join = 1; subflow->request_bkup = !!(flags & MPTCP_PM_ADDR_FLAG_BACKUP); subflow->subflow_id = msk->subflow_id++; @@ -XXX,XX +XXX,XX @@ static void subflow_ulp_clone(const struct request_sock *req, new_ctx->fully_established = 1; new_ctx->remote_key_valid = 1; new_ctx->backup = subflow_req->backup; - new_ctx->remote_id = subflow_req->remote_id; + WRITE_ONCE(new_ctx->remote_id, subflow_req->remote_id); new_ctx->token = subflow_req->token; new_ctx->thmac = subflow_req->thmac; -- 2.43.0 Fullmesh endpoints could end-up unexpectedly generating duplicate subflows - same local and remote addresses - when multiple incoming ADD_ADDR are processed before the PM creates the subflow for the local endpoints. Address the issue explicitly checking for duplicates at subflow creation time. To avoid a quadratic computational complexity, track the unavailable remote address ids in a temporary bitmap and initialize such bitmap with the remote ids of all the existing subflows matching the local address currently processed. The above allows additionally replacing the existing code checking for duplicate entry in the current set with a simple bit test operation. Fixes: 2843ff6f36db ("mptcp: remote addresses fullmesh") Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/435 Signed-off-by: Paolo Abeni <pabeni@redhat.com> --- Note that there is no problem for the opposite event sequence. --- net/mptcp/pm_netlink.c | 33 ++++++++++++++++++--------------- 1 file changed, 18 insertions(+), 15 deletions(-) diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -XXX,XX +XXX,XX @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk) } } -static bool lookup_address_in_vec(const struct mptcp_addr_info *addrs, unsigned int nr, - const struct mptcp_addr_info *addr) -{ - int i; - - for (i = 0; i < nr; i++) { - if (addrs[i].id == addr->id) - return true; - } - - return false; -} - /* Fill all the remote addresses into the array addrs[], * and return the array size. */ @@ -XXX,XX +XXX,XX @@ static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, msk->pm.subflows++; addrs[i++] = remote; } else { + DECLARE_BITMAP(unavail_id, MPTCP_PM_MAX_ADDR_ID + 1); + + /* Forbit creation of new subflows matching existing + * ones, possibly already created by incoming ADD_ADDR + */ + bitmap_zero(unavail_id, MPTCP_PM_MAX_ADDR_ID + 1); + mptcp_for_each_subflow(msk, subflow) + if (READ_ONCE(subflow->local_id) == local->id) + __set_bit(subflow->remote_id, unavail_id); + mptcp_for_each_subflow(msk, subflow) { ssk = mptcp_subflow_tcp_sock(subflow); remote_address((struct sock_common *)ssk, &addrs[i]); @@ -XXX,XX +XXX,XX @@ static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, if (deny_id0 && !addrs[i].id) continue; + if (test_bit(addrs[i].id, unavail_id)) + continue; + if (!mptcp_pm_addr_families_match(sk, local, &addrs[i])) continue; - if (!lookup_address_in_vec(addrs, i, &addrs[i]) && - msk->pm.subflows < subflows_max) { + if (msk->pm.subflows < subflows_max) { + /* forbit creating multiple address towards + * this id + */ + __set_bit(addrs[i].id, unavail_id); msk->pm.subflows++; i++; } -- 2.43.0 From: Paolo Abeni <pabeni@redhat.com> Fullmesh endpoints could end-up unexpectedly generating duplicate subflows - same local and remote addresses - when multiple incoming ADD_ADDR are processed before the PM creates the subflow for the local endpoints. Address the issue explicitly checking for duplicates at subflow creation time. To avoid a quadratic computational complexity, track the unavailable remote address ids in a temporary bitmap and initialize such bitmap with the remote ids of all the existing subflows matching the local address currently processed. The above allows additionally replacing the existing code checking for duplicate entry in the current set with a simple bit test operation. Fixes: 2843ff6f36db ("mptcp: remote addresses fullmesh") Cc: stable@vger.kernel.org Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/435 Signed-off-by: Paolo Abeni <pabeni@redhat.com> Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> (cherry picked from commit 045e9d812868a2d80b7a57b224ce8009444b7bbc) Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> --- Notes: - conflicts in pm_netlink.c because b9d69db87fb7 ("mptcp: let the in-kernel PM use mixed IPv4 and IPv6 addresses") is not in v6.1, and it introduced an extra check (mptcp_pm_addr_families_match()) in the modified context we don't need here. - we need the new 'local' parameter from b9d69db87fb7 ("mptcp: let the in-kernel PM use mixed IPv4 and IPv6 addresses"). --- net/mptcp/pm_netlink.c | 36 +++++++++++++++++++----------------- 1 file changed, 19 insertions(+), 17 deletions(-) diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index XXXXXXX..XXXXXXX 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -XXX,XX +XXX,XX @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk) } } -static bool lookup_address_in_vec(const struct mptcp_addr_info *addrs, unsigned int nr, - const struct mptcp_addr_info *addr) -{ - int i; - - for (i = 0; i < nr; i++) { - if (addrs[i].id == addr->id) - return true; - } - - return false; -} - /* Fill all the remote addresses into the array addrs[], * and return the array size. */ -static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, bool fullmesh, +static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, + struct mptcp_addr_info *local, + bool fullmesh, struct mptcp_addr_info *addrs) { bool deny_id0 = READ_ONCE(msk->pm.remote_deny_join_id0); @@ -XXX,XX +XXX,XX @@ static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, bool fullm msk->pm.subflows++; addrs[i++] = remote; } else { + DECLARE_BITMAP(unavail_id, MPTCP_PM_MAX_ADDR_ID + 1); + + /* Forbid creation of new subflows matching existing + * ones, possibly already created by incoming ADD_ADDR + */ + bitmap_zero(unavail_id, MPTCP_PM_MAX_ADDR_ID + 1); + mptcp_for_each_subflow(msk, subflow) + if (READ_ONCE(subflow->local_id) == local->id) + __set_bit(subflow->remote_id, unavail_id); + mptcp_for_each_subflow(msk, subflow) { ssk = mptcp_subflow_tcp_sock(subflow); remote_address((struct sock_common *)ssk, &addrs[i]); @@ -XXX,XX +XXX,XX @@ static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, bool fullm if (deny_id0 && !addrs[i].id) continue; - if (!lookup_address_in_vec(addrs, i, &addrs[i]) && - msk->pm.subflows < subflows_max) { + if (msk->pm.subflows < subflows_max) { + /* forbid creating multiple address towards + * this id + */ + __set_bit(addrs[i].id, unavail_id); msk->pm.subflows++; i++; } @@ -XXX,XX +XXX,XX @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) fullmesh = !!(local->flags & MPTCP_PM_ADDR_FLAG_FULLMESH); msk->pm.local_addr_used++; - nr = fill_remote_addresses_vec(msk, fullmesh, addrs); + nr = fill_remote_addresses_vec(msk, &local->addr, fullmesh, addrs); if (nr) __clear_bit(local->addr.id, msk->pm.id_avail_bitmap); spin_unlock_bh(&msk->pm.lock); -- 2.43.0