From 7ff57e98fb78ad94edafbdc7435f2d745e9e6bb5 Mon Sep 17 00:00:00 2001 From: "Fabio M. De Francesco" Date: Wed, 23 Feb 2022 11:02:52 +0100 Subject: net/smc: Use a mutex for locking "struct smc_pnettable" smc_pnetid_by_table_ib() uses read_lock() and then it calls smc_pnet_apply_ib() which, in turn, calls mutex_lock(&smc_ib_devices.mutex). read_lock() disables preemption. Therefore, the code acquires a mutex while in atomic context and it leads to a SAC bug. Fix this bug by replacing the rwlock with a mutex. Reported-and-tested-by: syzbot+4f322a6d84e991c38775@syzkaller.appspotmail.com Fixes: 64e28b52c7a6 ("net/smc: add pnet table namespace support") Confirmed-by: Tony Lu Signed-off-by: Fabio M. De Francesco Acked-by: Karsten Graul Link: https://lore.kernel.org/r/20220223100252.22562-1-fmdefrancesco@gmail.com Signed-off-by: Jakub Kicinski --- net/smc/smc_pnet.c | 42 +++++++++++++++++++++--------------------- net/smc/smc_pnet.h | 2 +- 2 files changed, 22 insertions(+), 22 deletions(-) (limited to 'net/smc') diff --git a/net/smc/smc_pnet.c b/net/smc/smc_pnet.c index 0599246c0376..29f0a559d884 100644 --- a/net/smc/smc_pnet.c +++ b/net/smc/smc_pnet.c @@ -113,7 +113,7 @@ static int smc_pnet_remove_by_pnetid(struct net *net, char *pnet_name) pnettable = &sn->pnettable; /* remove table entry */ - write_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry_safe(pnetelem, tmp_pe, &pnettable->pnetlist, list) { if (!pnet_name || @@ -131,7 +131,7 @@ static int smc_pnet_remove_by_pnetid(struct net *net, char *pnet_name) rc = 0; } } - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); /* if this is not the initial namespace, stop here */ if (net != &init_net) @@ -192,7 +192,7 @@ static int smc_pnet_add_by_ndev(struct net_device *ndev) sn = net_generic(net, smc_net_id); pnettable = &sn->pnettable; - write_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry_safe(pnetelem, tmp_pe, &pnettable->pnetlist, list) { if (pnetelem->type == SMC_PNET_ETH && !pnetelem->ndev && !strncmp(pnetelem->eth_name, ndev->name, IFNAMSIZ)) { @@ -206,7 +206,7 @@ static int smc_pnet_add_by_ndev(struct net_device *ndev) break; } } - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return rc; } @@ -224,7 +224,7 @@ static int smc_pnet_remove_by_ndev(struct net_device *ndev) sn = net_generic(net, smc_net_id); pnettable = &sn->pnettable; - write_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry_safe(pnetelem, tmp_pe, &pnettable->pnetlist, list) { if (pnetelem->type == SMC_PNET_ETH && pnetelem->ndev == ndev) { dev_put_track(pnetelem->ndev, &pnetelem->dev_tracker); @@ -237,7 +237,7 @@ static int smc_pnet_remove_by_ndev(struct net_device *ndev) break; } } - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return rc; } @@ -370,7 +370,7 @@ static int smc_pnet_add_eth(struct smc_pnettable *pnettable, struct net *net, strncpy(new_pe->eth_name, eth_name, IFNAMSIZ); rc = -EEXIST; new_netdev = true; - write_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(tmp_pe, &pnettable->pnetlist, list) { if (tmp_pe->type == SMC_PNET_ETH && !strncmp(tmp_pe->eth_name, eth_name, IFNAMSIZ)) { @@ -385,9 +385,9 @@ static int smc_pnet_add_eth(struct smc_pnettable *pnettable, struct net *net, GFP_ATOMIC); } list_add_tail(&new_pe->list, &pnettable->pnetlist); - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); } else { - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); kfree(new_pe); goto out_put; } @@ -448,7 +448,7 @@ static int smc_pnet_add_ib(struct smc_pnettable *pnettable, char *ib_name, new_pe->ib_port = ib_port; new_ibdev = true; - write_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(tmp_pe, &pnettable->pnetlist, list) { if (tmp_pe->type == SMC_PNET_IB && !strncmp(tmp_pe->ib_name, ib_name, IB_DEVICE_NAME_MAX)) { @@ -458,9 +458,9 @@ static int smc_pnet_add_ib(struct smc_pnettable *pnettable, char *ib_name, } if (new_ibdev) { list_add_tail(&new_pe->list, &pnettable->pnetlist); - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); } else { - write_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); kfree(new_pe); } return (new_ibdev) ? 0 : -EEXIST; @@ -605,7 +605,7 @@ static int _smc_pnet_dump(struct net *net, struct sk_buff *skb, u32 portid, pnettable = &sn->pnettable; /* dump pnettable entries */ - read_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(pnetelem, &pnettable->pnetlist, list) { if (pnetid && !smc_pnet_match(pnetelem->pnet_name, pnetid)) continue; @@ -620,7 +620,7 @@ static int _smc_pnet_dump(struct net *net, struct sk_buff *skb, u32 portid, break; } } - read_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return idx; } @@ -864,7 +864,7 @@ int smc_pnet_net_init(struct net *net) struct smc_pnetids_ndev *pnetids_ndev = &sn->pnetids_ndev; INIT_LIST_HEAD(&pnettable->pnetlist); - rwlock_init(&pnettable->lock); + mutex_init(&pnettable->lock); INIT_LIST_HEAD(&pnetids_ndev->list); rwlock_init(&pnetids_ndev->lock); @@ -944,7 +944,7 @@ static int smc_pnet_find_ndev_pnetid_by_table(struct net_device *ndev, sn = net_generic(net, smc_net_id); pnettable = &sn->pnettable; - read_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(pnetelem, &pnettable->pnetlist, list) { if (pnetelem->type == SMC_PNET_ETH && ndev == pnetelem->ndev) { /* get pnetid of netdev device */ @@ -953,7 +953,7 @@ static int smc_pnet_find_ndev_pnetid_by_table(struct net_device *ndev, break; } } - read_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return rc; } @@ -1156,7 +1156,7 @@ int smc_pnetid_by_table_ib(struct smc_ib_device *smcibdev, u8 ib_port) sn = net_generic(&init_net, smc_net_id); pnettable = &sn->pnettable; - read_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(tmp_pe, &pnettable->pnetlist, list) { if (tmp_pe->type == SMC_PNET_IB && !strncmp(tmp_pe->ib_name, ib_name, IB_DEVICE_NAME_MAX) && @@ -1166,7 +1166,7 @@ int smc_pnetid_by_table_ib(struct smc_ib_device *smcibdev, u8 ib_port) break; } } - read_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return rc; } @@ -1185,7 +1185,7 @@ int smc_pnetid_by_table_smcd(struct smcd_dev *smcddev) sn = net_generic(&init_net, smc_net_id); pnettable = &sn->pnettable; - read_lock(&pnettable->lock); + mutex_lock(&pnettable->lock); list_for_each_entry(tmp_pe, &pnettable->pnetlist, list) { if (tmp_pe->type == SMC_PNET_IB && !strncmp(tmp_pe->ib_name, ib_name, IB_DEVICE_NAME_MAX)) { @@ -1194,7 +1194,7 @@ int smc_pnetid_by_table_smcd(struct smcd_dev *smcddev) break; } } - read_unlock(&pnettable->lock); + mutex_unlock(&pnettable->lock); return rc; } diff --git a/net/smc/smc_pnet.h b/net/smc/smc_pnet.h index 14039272f7e4..80a88eea4949 100644 --- a/net/smc/smc_pnet.h +++ b/net/smc/smc_pnet.h @@ -29,7 +29,7 @@ struct smc_link_group; * @pnetlist: List of PNETIDs */ struct smc_pnettable { - rwlock_t lock; + struct mutex lock; struct list_head pnetlist; }; -- cgit v1.2.3 From 9f1c50cf39167ff71dc5953a3234f3f6eeb8fcb5 Mon Sep 17 00:00:00 2001 From: "D. Wythe" Date: Thu, 24 Feb 2022 23:26:19 +0800 Subject: net/smc: fix connection leak There's a potential leak issue under following execution sequence : smc_release smc_connect_work if (sk->sk_state == SMC_INIT) send_clc_confirim tcp_abort(); ... sk.sk_state = SMC_ACTIVE smc_close_active switch(sk->sk_state) { ... case SMC_ACTIVE: smc_close_final() // then wait peer closed Unfortunately, tcp_abort() may discard CLC CONFIRM messages that are still in the tcp send buffer, in which case our connection token cannot be delivered to the server side, which means that we cannot get a passive close message at all. Therefore, it is impossible for the to be disconnected at all. This patch tries a very simple way to avoid this issue, once the state has changed to SMC_ACTIVE after tcp_abort(), we can actively abort the smc connection, considering that the state is SMC_INIT before tcp_abort(), abandoning the complete disconnection process should not cause too much problem. In fact, this problem may exist as long as the CLC CONFIRM message is not received by the server. Whether a timer should be added after smc_close_final() needs to be discussed in the future. But even so, this patch provides a faster release for connection in above case, it should also be valuable. Fixes: 39f41f367b08 ("net/smc: common release code for non-accepted sockets") Signed-off-by: D. Wythe Acked-by: Karsten Graul Signed-off-by: David S. Miller --- net/smc/af_smc.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'net/smc') diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 306d9e8cd1dd..81984c1c0e78 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -183,7 +183,7 @@ static int smc_release(struct socket *sock) { struct sock *sk = sock->sk; struct smc_sock *smc; - int rc = 0; + int old_state, rc = 0; if (!sk) goto out; @@ -191,8 +191,10 @@ static int smc_release(struct socket *sock) sock_hold(sk); /* sock_put below */ smc = smc_sk(sk); + old_state = sk->sk_state; + /* cleanup for a dangling non-blocking connect */ - if (smc->connect_nonblock && sk->sk_state == SMC_INIT) + if (smc->connect_nonblock && old_state == SMC_INIT) tcp_abort(smc->clcsock->sk, ECONNABORTED); if (cancel_work_sync(&smc->connect_work)) @@ -206,6 +208,10 @@ static int smc_release(struct socket *sock) else lock_sock(sk); + if (old_state == SMC_INIT && sk->sk_state == SMC_ACTIVE && + !smc->use_fallback) + smc_close_active_abort(smc); + rc = __smc_release(smc); /* detach socket */ -- cgit v1.2.3 From 4d08b7b57ece83a1c31c633a7e4e27f121157f9c Mon Sep 17 00:00:00 2001 From: Tony Lu Date: Fri, 25 Feb 2022 14:56:57 +0800 Subject: net/smc: Fix cleanup when register ULP fails This patch calls smc_ib_unregister_client() when tcp_register_ulp() fails, and make sure to clean it up. Fixes: d7cd421da9da ("net/smc: Introduce TCP ULP support") Signed-off-by: Tony Lu Signed-off-by: David S. Miller --- net/smc/af_smc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'net/smc') diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 81984c1c0e78..284befa90967 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -3087,12 +3087,14 @@ static int __init smc_init(void) rc = tcp_register_ulp(&smc_ulp_ops); if (rc) { pr_err("%s: tcp_ulp_register fails with %d\n", __func__, rc); - goto out_sock; + goto out_ib; } static_branch_enable(&tcp_have_smc); return 0; +out_ib: + smc_ib_unregister_client(); out_sock: sock_unregister(PF_SMC); out_proto6: -- cgit v1.2.3 From 0537f0a2151375dcf90c1bbfda6a0aaf57164e89 Mon Sep 17 00:00:00 2001 From: "D. Wythe" Date: Wed, 2 Mar 2022 21:25:11 +0800 Subject: net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error generated by client The main reason for this unexpected SMC_CLC_DECL_ERR_REGRMB in client dues to following execution sequence: Server Conn A: Server Conn B: Client Conn B: smc_lgr_unregister_conn smc_lgr_register_conn smc_clc_send_accept -> smc_rtoken_add smcr_buf_unuse -> Client Conn A: smc_rtoken_delete smc_lgr_unregister_conn() makes current link available to assigned to new incoming connection, while smcr_buf_unuse() has not executed yet, which means that smc_rtoken_add may fail because of insufficient rtoken_entry, reversing their execution order will avoid this problem. Fixes: 3e034725c0d8 ("net/smc: common functions for RMBs and send buffers") Signed-off-by: D. Wythe Signed-off-by: David S. Miller --- net/smc/smc_core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/smc') diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c index 29525d03b253..f8c9675f4af3 100644 --- a/net/smc/smc_core.c +++ b/net/smc/smc_core.c @@ -1161,8 +1161,8 @@ void smc_conn_free(struct smc_connection *conn) cancel_work_sync(&conn->abort_work); } if (!list_empty(&lgr->list)) { - smc_lgr_unregister_conn(conn); smc_buf_unuse(conn, lgr); /* allow buffer reuse */ + smc_lgr_unregister_conn(conn); } if (!lgr->conns_num) -- cgit v1.2.3 From 4940a1fdf31c39f0806ac831cde333134862030b Mon Sep 17 00:00:00 2001 From: "D. Wythe" Date: Wed, 2 Mar 2022 21:25:12 +0800 Subject: net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error cause by server The problem of SMC_CLC_DECL_ERR_REGRMB on the server is very clear. Based on the fact that whether a new SMC connection can be accepted or not depends on not only the limit of conn nums, but also the available entries of rtoken. Since the rtoken release is trigger by peer, while the conn nums is decrease by local, tons of thing can happen in this time difference. This only thing that needs to be mentioned is that now all connection creations are completely protected by smc_server_lgr_pending lock, it's enough to check only the available entries in rtokens_used_mask. Fixes: cd6851f30386 ("smc: remote memory buffers (RMBs)") Signed-off-by: D. Wythe Signed-off-by: David S. Miller --- net/smc/smc_core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net/smc') diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c index f8c9675f4af3..be7d704976ff 100644 --- a/net/smc/smc_core.c +++ b/net/smc/smc_core.c @@ -1864,7 +1864,8 @@ int smc_conn_create(struct smc_sock *smc, struct smc_init_info *ini) (ini->smcd_version == SMC_V2 || lgr->vlan_id == ini->vlan_id) && (role == SMC_CLNT || ini->is_smcd || - lgr->conns_num < SMC_RMBS_PER_LGR_MAX)) { + (lgr->conns_num < SMC_RMBS_PER_LGR_MAX && + !bitmap_full(lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX)))) { /* link group found */ ini->first_contact_local = 0; conn->lgr = lgr; -- cgit v1.2.3