From c1aca3080e382886e2e58e809787441984a2f89b Mon Sep 17 00:00:00 2001 From: Yan Yan Date: Tue, 18 Jan 2022 16:00:13 -0800 Subject: xfrm: Check if_id in xfrm_migrate This patch enables distinguishing SAs and SPs based on if_id during the xfrm_migrate flow. This ensures support for xfrm interfaces throughout the SA/SP lifecycle. When there are multiple existing SPs with the same direction, the same xfrm_selector and different endpoint addresses, xfrm_migrate might fail with ENODATA. Specifically, the code path for performing xfrm_migrate is: Stage 1: find policy to migrate with xfrm_migrate_policy_find(sel, dir, type, net) Stage 2: find and update state(s) with xfrm_migrate_state_find(mp, net) Stage 3: update endpoint address(es) of template(s) with xfrm_policy_migrate(pol, m, num_migrate) Currently "Stage 1" always returns the first xfrm_policy that matches, and "Stage 3" looks for the xfrm_tmpl that matches the old endpoint address. Thus if there are multiple xfrm_policy with same selector, direction, type and net, "Stage 1" might rertun a wrong xfrm_policy and "Stage 3" will fail with ENODATA because it cannot find a xfrm_tmpl with the matching endpoint address. The fix is to allow userspace to pass an if_id and add if_id to the matching rule in Stage 1 and Stage 2 since if_id is a unique ID for xfrm_policy and xfrm_state. For compatibility, if_id will only be checked if the attribute is set. Tested with additions to Android's kernel unit test suite: https://android-review.googlesource.com/c/kernel/tests/+/1668886 Signed-off-by: Yan Yan Signed-off-by: Steffen Klassert --- include/net/xfrm.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/net/xfrm.h b/include/net/xfrm.h index fdb41e8bb626..743dd1da506e 100644 --- a/include/net/xfrm.h +++ b/include/net/xfrm.h @@ -1681,14 +1681,15 @@ int km_migrate(const struct xfrm_selector *sel, u8 dir, u8 type, const struct xfrm_migrate *m, int num_bundles, const struct xfrm_kmaddress *k, const struct xfrm_encap_tmpl *encap); -struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *net); +struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *net, + u32 if_id); struct xfrm_state *xfrm_state_migrate(struct xfrm_state *x, struct xfrm_migrate *m, struct xfrm_encap_tmpl *encap); int xfrm_migrate(const struct xfrm_selector *sel, u8 dir, u8 type, struct xfrm_migrate *m, int num_bundles, struct xfrm_kmaddress *k, struct net *net, - struct xfrm_encap_tmpl *encap); + struct xfrm_encap_tmpl *encap, u32 if_id); #endif int km_new_mapping(struct xfrm_state *x, xfrm_address_t *ipaddr, __be16 sport); -- cgit v1.2.3 From a6d95c5a628a09be129f25d5663a7e9db8261f51 Mon Sep 17 00:00:00 2001 From: Jiri Bohac Date: Wed, 26 Jan 2022 16:00:18 +0100 Subject: Revert "xfrm: xfrm_state_mtu should return at least 1280 for ipv6" This reverts commit b515d2637276a3810d6595e10ab02c13bfd0b63a. Commit b515d2637276a3810d6595e10ab02c13bfd0b63a ("xfrm: xfrm_state_mtu should return at least 1280 for ipv6") in v5.14 breaks the TCP MSS calculation in ipsec transport mode, resulting complete stalls of TCP connections. This happens when the (P)MTU is 1280 or slighly larger. The desired formula for the MSS is: MSS = (MTU - ESP_overhead) - IP header - TCP header However, the above commit clamps the (MTU - ESP_overhead) to a minimum of 1280, turning the formula into MSS = max(MTU - ESP overhead, 1280) - IP header - TCP header With the (P)MTU near 1280, the calculated MSS is too large and the resulting TCP packets never make it to the destination because they are over the actual PMTU. The above commit also causes suboptimal double fragmentation in xfrm tunnel mode, as described in https://lore.kernel.org/netdev/20210429202529.codhwpc7w6kbudug@dwarf.suse.cz/ The original problem the above commit was trying to fix is now fixed by commit 6596a0229541270fb8d38d989f91b78838e5e9da ("xfrm: fix MTU regression"). Signed-off-by: Jiri Bohac Signed-off-by: Steffen Klassert --- include/net/xfrm.h | 1 - net/ipv4/esp4.c | 2 +- net/ipv6/esp6.c | 2 +- net/xfrm/xfrm_state.c | 14 ++------------ 4 files changed, 4 insertions(+), 15 deletions(-) (limited to 'include') diff --git a/include/net/xfrm.h b/include/net/xfrm.h index 743dd1da506e..76aa6f11a540 100644 --- a/include/net/xfrm.h +++ b/include/net/xfrm.h @@ -1568,7 +1568,6 @@ void xfrm_sad_getinfo(struct net *net, struct xfrmk_sadinfo *si); void xfrm_spd_getinfo(struct net *net, struct xfrmk_spdinfo *si); u32 xfrm_replay_seqhi(struct xfrm_state *x, __be32 net_seq); int xfrm_init_replay(struct xfrm_state *x); -u32 __xfrm_state_mtu(struct xfrm_state *x, int mtu); u32 xfrm_state_mtu(struct xfrm_state *x, int mtu); int __xfrm_init_state(struct xfrm_state *x, bool init_replay, bool offload); int xfrm_init_state(struct xfrm_state *x); diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c index 851f542928a3..e1b1d080e908 100644 --- a/net/ipv4/esp4.c +++ b/net/ipv4/esp4.c @@ -671,7 +671,7 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb) struct xfrm_dst *dst = (struct xfrm_dst *)skb_dst(skb); u32 padto; - padto = min(x->tfcpad, __xfrm_state_mtu(x, dst->child_mtu_cached)); + padto = min(x->tfcpad, xfrm_state_mtu(x, dst->child_mtu_cached)); if (skb->len < padto) esp.tfclen = padto - skb->len; } diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c index 8bb2c407b46b..7591160edce1 100644 --- a/net/ipv6/esp6.c +++ b/net/ipv6/esp6.c @@ -707,7 +707,7 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb) struct xfrm_dst *dst = (struct xfrm_dst *)skb_dst(skb); u32 padto; - padto = min(x->tfcpad, __xfrm_state_mtu(x, dst->child_mtu_cached)); + padto = min(x->tfcpad, xfrm_state_mtu(x, dst->child_mtu_cached)); if (skb->len < padto) esp.tfclen = padto - skb->len; } diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c index 1ba6fbfe8cdb..b749935152ba 100644 --- a/net/xfrm/xfrm_state.c +++ b/net/xfrm/xfrm_state.c @@ -2579,7 +2579,7 @@ void xfrm_state_delete_tunnel(struct xfrm_state *x) } EXPORT_SYMBOL(xfrm_state_delete_tunnel); -u32 __xfrm_state_mtu(struct xfrm_state *x, int mtu) +u32 xfrm_state_mtu(struct xfrm_state *x, int mtu) { const struct xfrm_type *type = READ_ONCE(x->type); struct crypto_aead *aead; @@ -2610,17 +2610,7 @@ u32 __xfrm_state_mtu(struct xfrm_state *x, int mtu) return ((mtu - x->props.header_len - crypto_aead_authsize(aead) - net_adj) & ~(blksize - 1)) + net_adj - 2; } -EXPORT_SYMBOL_GPL(__xfrm_state_mtu); - -u32 xfrm_state_mtu(struct xfrm_state *x, int mtu) -{ - mtu = __xfrm_state_mtu(x, mtu); - - if (x->props.family == AF_INET6 && mtu < IPV6_MIN_MTU) - return IPV6_MIN_MTU; - - return mtu; -} +EXPORT_SYMBOL_GPL(xfrm_state_mtu); int __xfrm_init_state(struct xfrm_state *x, bool init_replay, bool offload) { -- cgit v1.2.3 From 7c76ecd9c99b6e9a771d813ab1aa7fa428b3ade1 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 8 Feb 2022 16:14:32 +0200 Subject: xfrm: enforce validity of offload input flags struct xfrm_user_offload has flags variable that received user input, but kernel didn't check if valid bits were provided. It caused a situation where not sanitized input was forwarded directly to the drivers. For example, XFRM_OFFLOAD_IPV6 define that was exposed, was used by strongswan, but not implemented in the kernel at all. As a solution, check and sanitize input flags to forward XFRM_OFFLOAD_INBOUND to the drivers. Fixes: d77e38e612a0 ("xfrm: Add an IPsec hardware offloading API") Signed-off-by: Leon Romanovsky Signed-off-by: Steffen Klassert --- include/uapi/linux/xfrm.h | 6 ++++++ net/xfrm/xfrm_device.c | 6 +++++- 2 files changed, 11 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/linux/xfrm.h b/include/uapi/linux/xfrm.h index 4e29d7851890..65e13a099b1a 100644 --- a/include/uapi/linux/xfrm.h +++ b/include/uapi/linux/xfrm.h @@ -511,6 +511,12 @@ struct xfrm_user_offload { int ifindex; __u8 flags; }; +/* This flag was exposed without any kernel code that supporting it. + * Unfortunately, strongswan has the code that uses sets this flag, + * which makes impossible to reuse this bit. + * + * So leave it here to make sure that it won't be reused by mistake. + */ #define XFRM_OFFLOAD_IPV6 1 #define XFRM_OFFLOAD_INBOUND 2 diff --git a/net/xfrm/xfrm_device.c b/net/xfrm/xfrm_device.c index 3fa066419d37..39bce5d764de 100644 --- a/net/xfrm/xfrm_device.c +++ b/net/xfrm/xfrm_device.c @@ -223,6 +223,9 @@ int xfrm_dev_state_add(struct net *net, struct xfrm_state *x, if (x->encap || x->tfcpad) return -EINVAL; + if (xuo->flags & ~(XFRM_OFFLOAD_IPV6 | XFRM_OFFLOAD_INBOUND)) + return -EINVAL; + dev = dev_get_by_index(net, xuo->ifindex); if (!dev) { if (!(xuo->flags & XFRM_OFFLOAD_INBOUND)) { @@ -262,7 +265,8 @@ int xfrm_dev_state_add(struct net *net, struct xfrm_state *x, netdev_tracker_alloc(dev, &xso->dev_tracker, GFP_ATOMIC); xso->real_dev = dev; xso->num_exthdrs = 1; - xso->flags = xuo->flags; + /* Don't forward bit that is not implemented */ + xso->flags = xuo->flags & ~XFRM_OFFLOAD_IPV6; err = dev->xfrmdev_ops->xdo_dev_state_add(x); if (err) { -- cgit v1.2.3 From 29fb608396d6a62c1b85acc421ad7a4399085b9f Mon Sep 17 00:00:00 2001 From: Luiz Augusto von Dentz Date: Mon, 14 Feb 2022 17:59:38 -0800 Subject: Bluetooth: Fix bt_skb_sendmmsg not allocating partial chunks Since bt_skb_sendmmsg can be used with the likes of SOCK_STREAM it shall return the partial chunks it could allocate instead of freeing everything as otherwise it can cause problems like bellow. Fixes: 81be03e026dc ("Bluetooth: RFCOMM: Replace use of memcpy_from_msg with bt_skb_sendmmsg") Reported-by: Paul Menzel Link: https://lore.kernel.org/r/d7206e12-1b99-c3be-84f4-df22af427ef5@molgen.mpg.de BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=215594 Signed-off-by: Luiz Augusto von Dentz Tested-by: Paul Menzel (Nokia N9 (MeeGo/Harmattan) Signed-off-by: Marcel Holtmann --- include/net/bluetooth/bluetooth.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'include') diff --git a/include/net/bluetooth/bluetooth.h b/include/net/bluetooth/bluetooth.h index 4b3d0b16c185..a647e5fabdbd 100644 --- a/include/net/bluetooth/bluetooth.h +++ b/include/net/bluetooth/bluetooth.h @@ -506,8 +506,7 @@ static inline struct sk_buff *bt_skb_sendmmsg(struct sock *sk, tmp = bt_skb_sendmsg(sk, msg, len, mtu, headroom, tailroom); if (IS_ERR(tmp)) { - kfree_skb(skb); - return tmp; + return skb; } len -= tmp->len; -- cgit v1.2.3 From a56a1138cbd85e4d565356199d60e1cb94e5a77a Mon Sep 17 00:00:00 2001 From: Luiz Augusto von Dentz Date: Thu, 17 Feb 2022 13:10:38 -0800 Subject: Bluetooth: hci_sync: Fix not using conn_timeout When using hci_le_create_conn_sync it shall wait for the conn_timeout since the connection complete may take longer than just 2 seconds. Also fix the masking of HCI_EV_LE_ENHANCED_CONN_COMPLETE and HCI_EV_LE_CONN_COMPLETE so they are never both set so we can predict which one the controller will use in case of HCI_OP_LE_CREATE_CONN. Fixes: 6cd29ec6ae5e3 ("Bluetooth: hci_sync: Wait for proper events when connecting LE") Signed-off-by: Luiz Augusto von Dentz Signed-off-by: Marcel Holtmann --- include/net/bluetooth/hci_core.h | 8 ++++++++ net/bluetooth/hci_sync.c | 21 +++++++++++++++------ 2 files changed, 23 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h index 586f69d084a2..e336e9c1dda4 100644 --- a/include/net/bluetooth/hci_core.h +++ b/include/net/bluetooth/hci_core.h @@ -1489,6 +1489,14 @@ void hci_conn_del_sysfs(struct hci_conn *conn); /* Extended advertising support */ #define ext_adv_capable(dev) (((dev)->le_features[1] & HCI_LE_EXT_ADV)) +/* BLUETOOTH CORE SPECIFICATION Version 5.3 | Vol 4, Part E page 1789: + * + * C24: Mandatory if the LE Controller supports Connection State and either + * LE Feature (LL Privacy) or LE Feature (Extended Advertising) is supported + */ +#define use_enhanced_conn_complete(dev) (ll_privacy_capable(dev) || \ + ext_adv_capable(dev)) + /* ----- HCI protocols ----- */ #define HCI_PROTO_DEFER 0x01 diff --git a/net/bluetooth/hci_sync.c b/net/bluetooth/hci_sync.c index 9d8490570b42..9ba2a1a7d481 100644 --- a/net/bluetooth/hci_sync.c +++ b/net/bluetooth/hci_sync.c @@ -3265,10 +3265,10 @@ static int hci_le_set_event_mask_sync(struct hci_dev *hdev) if (hdev->le_features[0] & HCI_LE_DATA_LEN_EXT) events[0] |= 0x40; /* LE Data Length Change */ - /* If the controller supports LL Privacy feature, enable - * the corresponding event. + /* If the controller supports LL Privacy feature or LE Extended Adv, + * enable the corresponding event. */ - if (hdev->le_features[0] & HCI_LE_LL_PRIVACY) + if (use_enhanced_conn_complete(hdev)) events[1] |= 0x02; /* LE Enhanced Connection Complete */ /* If the controller supports Extended Scanner Filter @@ -5188,7 +5188,7 @@ int hci_le_ext_create_conn_sync(struct hci_dev *hdev, struct hci_conn *conn, return __hci_cmd_sync_status_sk(hdev, HCI_OP_LE_EXT_CREATE_CONN, plen, data, HCI_EV_LE_ENHANCED_CONN_COMPLETE, - HCI_CMD_TIMEOUT, NULL); + conn->conn_timeout, NULL); } int hci_le_create_conn_sync(struct hci_dev *hdev, struct hci_conn *conn) @@ -5273,9 +5273,18 @@ int hci_le_create_conn_sync(struct hci_dev *hdev, struct hci_conn *conn) cp.min_ce_len = cpu_to_le16(0x0000); cp.max_ce_len = cpu_to_le16(0x0000); + /* BLUETOOTH CORE SPECIFICATION Version 5.3 | Vol 4, Part E page 2261: + * + * If this event is unmasked and the HCI_LE_Connection_Complete event + * is unmasked, only the HCI_LE_Enhanced_Connection_Complete event is + * sent when a new connection has been created. + */ err = __hci_cmd_sync_status_sk(hdev, HCI_OP_LE_CREATE_CONN, - sizeof(cp), &cp, HCI_EV_LE_CONN_COMPLETE, - HCI_CMD_TIMEOUT, NULL); + sizeof(cp), &cp, + use_enhanced_conn_complete(hdev) ? + HCI_EV_LE_ENHANCED_CONN_COMPLETE : + HCI_EV_LE_CONN_COMPLETE, + conn->conn_timeout, NULL); done: /* Re-enable advertising after the connection attempt is finished. */ -- cgit v1.2.3 From 17a8f31bba7bac8cce4bd12bab50697da96e7710 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 28 Feb 2022 04:18:05 +0100 Subject: netfilter: egress: silence egress hook lockdep splats Netfilter assumes its called with rcu_read_lock held, but in egress hook case it may be called with BH readlock. This triggers lockdep splat. In order to avoid to change all rcu_dereference() to rcu_dereference_check(..., rcu_read_lock_bh_held()), wrap nf_hook_slow with read lock/unlock pair. Reported-by: Eric Dumazet Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter_netdev.h | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'include') diff --git a/include/linux/netfilter_netdev.h b/include/linux/netfilter_netdev.h index b4dd96e4dc8d..e6487a691136 100644 --- a/include/linux/netfilter_netdev.h +++ b/include/linux/netfilter_netdev.h @@ -101,7 +101,11 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc, nf_hook_state_init(&state, NF_NETDEV_EGRESS, NFPROTO_NETDEV, dev, NULL, NULL, dev_net(dev), NULL); + + /* nf assumes rcu_read_lock, not just read_lock_bh */ + rcu_read_lock(); ret = nf_hook_slow(skb, &state, e, 0); + rcu_read_unlock(); if (ret == 1) { return skb; -- cgit v1.2.3 From 50bb467c9e76743fbc8441d29113cdad62dbc4fe Mon Sep 17 00:00:00 2001 From: Ben Dooks Date: Fri, 18 Feb 2022 09:38:58 +0000 Subject: rfkill: define rfill_soft_blocked() if !RFKILL If CONFIG_RFKILL is not set, the Intel WiFi driver will not build the iw_mvm driver part due to the missing rfill_soft_blocked() call. Adding a inline declaration of rfill_soft_blocked() if CONFIG_RFKILL=n fixes the following error: drivers/net/wireless/intel/iwlwifi/mvm/mvm.h: In function 'iwl_mvm_mei_set_sw_rfkill_state': drivers/net/wireless/intel/iwlwifi/mvm/mvm.h:2215:38: error: implicit declaration of function 'rfkill_soft_blocked'; did you mean 'rfkill_blocked'? [-Werror=implicit-function-declaration] 2215 | mvm->hw_registered ? rfkill_soft_blocked(mvm->hw->wiphy->rfkill) : false; | ^~~~~~~~~~~~~~~~~~~ | rfkill_blocked Signed-off-by: Ben Dooks Reported-by: Neill Whillans Fixes: 5bc9a9dd7535 ("rfkill: allow to get the software rfkill state") Link: https://lore.kernel.org/r/20220218093858.1245677-1-ben.dooks@codethink.co.uk Signed-off-by: Johannes Berg --- include/linux/rfkill.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include') diff --git a/include/linux/rfkill.h b/include/linux/rfkill.h index c35f3962dc4f..373003ace639 100644 --- a/include/linux/rfkill.h +++ b/include/linux/rfkill.h @@ -308,6 +308,11 @@ static inline bool rfkill_blocked(struct rfkill *rfkill) return false; } +static inline bool rfkill_soft_blocked(struct rfkill *rfkill) +{ + return false; +} + static inline enum rfkill_type rfkill_find_type(const char *name) { return RFKILL_TYPE_ALL; -- cgit v1.2.3 From c3873070247d9e3c7a6b0cf9bf9b45e8018427b1 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 28 Feb 2022 06:22:22 +0100 Subject: netfilter: nf_queue: fix possible use-after-free Eric Dumazet says: The sock_hold() side seems suspect, because there is no guarantee that sk_refcnt is not already 0. On failure, we cannot queue the packet and need to indicate an error. The packet will be dropped by the caller. v2: split skb prefetch hunk into separate change Fixes: 271b72c7fa82c ("udp: RCU handling for Unicast packets.") Reported-by: Eric Dumazet Reviewed-by: Eric Dumazet Signed-off-by: Florian Westphal --- include/net/netfilter/nf_queue.h | 2 +- net/netfilter/nf_queue.c | 13 +++++++++---- net/netfilter/nfnetlink_queue.c | 12 +++++++++--- 3 files changed, 19 insertions(+), 8 deletions(-) (limited to 'include') diff --git a/include/net/netfilter/nf_queue.h b/include/net/netfilter/nf_queue.h index 9eed51e920e8..980daa6e1e3a 100644 --- a/include/net/netfilter/nf_queue.h +++ b/include/net/netfilter/nf_queue.h @@ -37,7 +37,7 @@ void nf_register_queue_handler(const struct nf_queue_handler *qh); void nf_unregister_queue_handler(void); void nf_reinject(struct nf_queue_entry *entry, unsigned int verdict); -void nf_queue_entry_get_refs(struct nf_queue_entry *entry); +bool nf_queue_entry_get_refs(struct nf_queue_entry *entry); void nf_queue_entry_free(struct nf_queue_entry *entry); static inline void init_hashrandom(u32 *jhash_initval) diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c index 5ab0680db445..e39549c55945 100644 --- a/net/netfilter/nf_queue.c +++ b/net/netfilter/nf_queue.c @@ -96,19 +96,21 @@ static void __nf_queue_entry_init_physdevs(struct nf_queue_entry *entry) } /* Bump dev refs so they don't vanish while packet is out */ -void nf_queue_entry_get_refs(struct nf_queue_entry *entry) +bool nf_queue_entry_get_refs(struct nf_queue_entry *entry) { struct nf_hook_state *state = &entry->state; + if (state->sk && !refcount_inc_not_zero(&state->sk->sk_refcnt)) + return false; + dev_hold(state->in); dev_hold(state->out); - if (state->sk) - sock_hold(state->sk); #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) dev_hold(entry->physin); dev_hold(entry->physout); #endif + return true; } EXPORT_SYMBOL_GPL(nf_queue_entry_get_refs); @@ -196,7 +198,10 @@ static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, __nf_queue_entry_init_physdevs(entry); - nf_queue_entry_get_refs(entry); + if (!nf_queue_entry_get_refs(entry)) { + kfree(entry); + return -ENOTCONN; + } switch (entry->state.pf) { case AF_INET: diff --git a/net/netfilter/nfnetlink_queue.c b/net/netfilter/nfnetlink_queue.c index ea2d9c2a44cf..64a6acb6aeae 100644 --- a/net/netfilter/nfnetlink_queue.c +++ b/net/netfilter/nfnetlink_queue.c @@ -710,9 +710,15 @@ static struct nf_queue_entry * nf_queue_entry_dup(struct nf_queue_entry *e) { struct nf_queue_entry *entry = kmemdup(e, e->size, GFP_ATOMIC); - if (entry) - nf_queue_entry_get_refs(entry); - return entry; + + if (!entry) + return NULL; + + if (nf_queue_entry_get_refs(entry)) + return entry; + + kfree(entry); + return NULL; } #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) -- cgit v1.2.3 From db6140e5e35a48405e669353bd54042c1d4c3841 Mon Sep 17 00:00:00 2001 From: Paul Blakey Date: Mon, 28 Feb 2022 11:23:49 +0200 Subject: net/sched: act_ct: Fix flow table lookup failure with no originating ifindex After cited commit optimizted hw insertion, flow table entries are populated with ifindex information which was intended to only be used for HW offload. This tuple ifindex is hashed in the flow table key, so it must be filled for lookup to be successful. But tuple ifindex is only relevant for the netfilter flowtables (nft), so it's not filled in act_ct flow table lookup, resulting in lookup failure, and no SW offload and no offload teardown for TCP connection FIN/RST packets. To fix this, add new tc ifindex field to tuple, which will only be used for offloading, not for lookup, as it will not be part of the tuple hash. Fixes: 9795ded7f924 ("net/sched: act_ct: Fill offloading tuple iifidx") Signed-off-by: Paul Blakey Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_flow_table.h | 6 +++++- net/netfilter/nf_flow_table_offload.c | 6 +++++- net/sched/act_ct.c | 13 +++++++++---- 3 files changed, 19 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/include/net/netfilter/nf_flow_table.h b/include/net/netfilter/nf_flow_table.h index a3647fadf1cc..bd59e950f4d6 100644 --- a/include/net/netfilter/nf_flow_table.h +++ b/include/net/netfilter/nf_flow_table.h @@ -96,6 +96,7 @@ enum flow_offload_xmit_type { FLOW_OFFLOAD_XMIT_NEIGH, FLOW_OFFLOAD_XMIT_XFRM, FLOW_OFFLOAD_XMIT_DIRECT, + FLOW_OFFLOAD_XMIT_TC, }; #define NF_FLOW_TABLE_ENCAP_MAX 2 @@ -127,7 +128,7 @@ struct flow_offload_tuple { struct { } __hash; u8 dir:2, - xmit_type:2, + xmit_type:3, encap_num:2, in_vlan_ingress:2; u16 mtu; @@ -142,6 +143,9 @@ struct flow_offload_tuple { u8 h_source[ETH_ALEN]; u8 h_dest[ETH_ALEN]; } out; + struct { + u32 iifidx; + } tc; }; }; diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index b561e0a44a45..fc4265acd9c4 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -110,7 +110,11 @@ static int nf_flow_rule_match(struct nf_flow_match *match, nf_flow_rule_lwt_match(match, tun_info); } - key->meta.ingress_ifindex = tuple->iifidx; + if (tuple->xmit_type == FLOW_OFFLOAD_XMIT_TC) + key->meta.ingress_ifindex = tuple->tc.iifidx; + else + key->meta.ingress_ifindex = tuple->iifidx; + mask->meta.ingress_ifindex = 0xffffffff; if (tuple->encap_num > 0 && !(tuple->in_vlan_ingress & BIT(0)) && diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 33e70d60f0bf..ec19f625863a 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -361,6 +361,13 @@ static void tcf_ct_flow_table_put(struct tcf_ct_params *params) } } +static void tcf_ct_flow_tc_ifidx(struct flow_offload *entry, + struct nf_conn_act_ct_ext *act_ct_ext, u8 dir) +{ + entry->tuplehash[dir].tuple.xmit_type = FLOW_OFFLOAD_XMIT_TC; + entry->tuplehash[dir].tuple.tc.iifidx = act_ct_ext->ifindex[dir]; +} + static void tcf_ct_flow_table_add(struct tcf_ct_flow_table *ct_ft, struct nf_conn *ct, bool tcp) @@ -385,10 +392,8 @@ static void tcf_ct_flow_table_add(struct tcf_ct_flow_table *ct_ft, act_ct_ext = nf_conn_act_ct_ext_find(ct); if (act_ct_ext) { - entry->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.iifidx = - act_ct_ext->ifindex[IP_CT_DIR_ORIGINAL]; - entry->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.iifidx = - act_ct_ext->ifindex[IP_CT_DIR_REPLY]; + tcf_ct_flow_tc_ifidx(entry, act_ct_ext, FLOW_OFFLOAD_DIR_ORIGINAL); + tcf_ct_flow_tc_ifidx(entry, act_ct_ext, FLOW_OFFLOAD_DIR_REPLY); } err = flow_offload_add(&ct_ft->nf_ft, entry); -- cgit v1.2.3 From 2d3916f3189172d5c69d33065c3c21119fe539fc Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 3 Mar 2022 09:37:28 -0800 Subject: ipv6: fix skb drops in igmp6_event_query() and igmp6_event_report() While investigating on why a synchronize_net() has been added recently in ipv6_mc_down(), I found that igmp6_event_query() and igmp6_event_report() might drop skbs in some cases. Discussion about removing synchronize_net() from ipv6_mc_down() will happen in a different thread. Fixes: f185de28d9ae ("mld: add new workqueues for process mld events") Signed-off-by: Eric Dumazet Cc: Taehee Yoo Cc: Cong Wang Cc: David Ahern Link: https://lore.kernel.org/r/20220303173728.937869-1-eric.dumazet@gmail.com Signed-off-by: Jakub Kicinski --- include/net/ndisc.h | 4 ++-- net/ipv6/mcast.c | 32 ++++++++++++-------------------- 2 files changed, 14 insertions(+), 22 deletions(-) (limited to 'include') diff --git a/include/net/ndisc.h b/include/net/ndisc.h index 53cb8de0e589..47ffb360ddfa 100644 --- a/include/net/ndisc.h +++ b/include/net/ndisc.h @@ -475,9 +475,9 @@ int igmp6_late_init(void); void igmp6_cleanup(void); void igmp6_late_cleanup(void); -int igmp6_event_query(struct sk_buff *skb); +void igmp6_event_query(struct sk_buff *skb); -int igmp6_event_report(struct sk_buff *skb); +void igmp6_event_report(struct sk_buff *skb); #ifdef CONFIG_SYSCTL diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c index a8861db52c18..909f937befd7 100644 --- a/net/ipv6/mcast.c +++ b/net/ipv6/mcast.c @@ -1371,27 +1371,23 @@ static void mld_process_v2(struct inet6_dev *idev, struct mld2_query *mld, } /* called with rcu_read_lock() */ -int igmp6_event_query(struct sk_buff *skb) +void igmp6_event_query(struct sk_buff *skb) { struct inet6_dev *idev = __in6_dev_get(skb->dev); - if (!idev) - return -EINVAL; - - if (idev->dead) { - kfree_skb(skb); - return -ENODEV; - } + if (!idev || idev->dead) + goto out; spin_lock_bh(&idev->mc_query_lock); if (skb_queue_len(&idev->mc_query_queue) < MLD_MAX_SKBS) { __skb_queue_tail(&idev->mc_query_queue, skb); if (!mod_delayed_work(mld_wq, &idev->mc_query_work, 0)) in6_dev_hold(idev); + skb = NULL; } spin_unlock_bh(&idev->mc_query_lock); - - return 0; +out: + kfree_skb(skb); } static void __mld_query_work(struct sk_buff *skb) @@ -1542,27 +1538,23 @@ static void mld_query_work(struct work_struct *work) } /* called with rcu_read_lock() */ -int igmp6_event_report(struct sk_buff *skb) +void igmp6_event_report(struct sk_buff *skb) { struct inet6_dev *idev = __in6_dev_get(skb->dev); - if (!idev) - return -EINVAL; - - if (idev->dead) { - kfree_skb(skb); - return -ENODEV; - } + if (!idev || idev->dead) + goto out; spin_lock_bh(&idev->mc_report_lock); if (skb_queue_len(&idev->mc_report_queue) < MLD_MAX_SKBS) { __skb_queue_tail(&idev->mc_report_queue, skb); if (!mod_delayed_work(mld_wq, &idev->mc_report_work, 0)) in6_dev_hold(idev); + skb = NULL; } spin_unlock_bh(&idev->mc_report_lock); - - return 0; +out: + kfree_skb(skb); } static void __mld_report_work(struct sk_buff *skb) -- cgit v1.2.3