- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6 0/2] CVE-2026-23276
by Li Xiasong 02 Apr '26

02 Apr '26

Eric Dumazet (1): net: prevent NULL deref in ip[6]tunnel_xmit() Weiming Shi (1): net: add xmit recursion limit to tunnel xmit functions include/net/ip6_tunnel.h | 14 ++++++++++++++ include/net/ip_tunnels.h | 7 +++++++ net/ipv4/ip_tunnel_core.c | 15 +++++++++++++++ 3 files changed, 36 insertions(+) -- 2.34.1

2 3

[PATCH OLK-6.6] tcp: secure_seq: add back ports to TS offset
by Li Xiasong 02 Apr '26

02 Apr '26

From: Eric Dumazet <edumazet(a)google.com> mainline inclusion from mainline-v7.0-rc3 commit 165573e41f2f66ef98940cf65f838b2cb575d9d1 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/13867 CVE: CVE-2026-23247 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- This reverts 28ee1b746f49 ("secure_seq: downgrade to per-host timestamp offsets") tcp_tw_recycle went away in 2017. Zhouyan Deng reported off-path TCP source port leakage via SYN cookie side-channel that can be fixed in multiple ways. One of them is to bring back TCP ports in TS offset randomization. As a bonus, we perform a single siphash() computation to provide both an ISN and a TS offset. Fixes: 28ee1b746f49 ("secure_seq: downgrade to per-host timestamp offsets") Reported-by: Zhouyan Deng <dengzhouyan_nwpu(a)163.com> Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: Kuniyuki Iwashima <kuniyu(a)google.com> Acked-by: Florian Westphal <fw(a)strlen.de> Link: https://patch.msgid.link/20260302205527.1982836-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Conflicts: include/net/secure_seq.h include/net/tcp.h net/core/secure_seq.c net/ipv4/syncookies.c net/ipv4/tcp_input.c net/ipv4/tcp_ipv4.c net/ipv6/syncookies.c net/ipv6/tcp_ipv6.c [conflicts due to merged bf8317ce3910 ("net: use __GENKSYMS__ to revert the kabi change") and not merge 2a63dd0edf38 ("net: Retire DCCP socket.") and 6dc4c2526f6d ("tcp: use EXPORT_IPV6_MOD[_GPL]()") and 8e7bab6b9652 ("tcp: Factorise cookie-independent fields initialisation in cookie_v[46]_check()") and cdbab6236605 ("tcp: fix fastopen code vs usec TS") and 449f68f8fffa ("net: Convert proto callbacks from sockaddr to sockaddr_unsized").] Signed-off-by: Li Xiasong <lixiasong1(a)huawei.com> --- include/net/secure_seq.h | 45 ++++++++++++++++++---- include/net/tcp.h | 6 ++- net/core/secure_seq.c | 80 +++++++++++++++------------------------- net/ipv4/syncookies.c | 11 ++++-- net/ipv4/tcp_input.c | 10 +++-- net/ipv4/tcp_ipv4.c | 37 +++++++++---------- net/ipv6/syncookies.c | 11 ++++-- net/ipv6/tcp_ipv6.c | 37 +++++++++---------- 8 files changed, 128 insertions(+), 109 deletions(-) diff --git a/include/net/secure_seq.h b/include/net/secure_seq.h index 21e7fa2a1813..e4ad681a2a12 100644 --- a/include/net/secure_seq.h +++ b/include/net/secure_seq.h @@ -5,20 +5,51 @@ #include <linux/types.h> struct net; +extern struct net init_net; + +union tcp_seq_and_ts_off { + struct { + u32 seq; + u32 ts_off; + }; + u64 hash64; +}; u64 secure_ipv4_port_ephemeral(__be32 saddr, __be32 daddr, __be16 dport); u64 secure_ipv6_port_ephemeral(const __be32 *saddr, const __be32 *daddr, __be16 dport); -u32 secure_tcp_seq(__be32 saddr, __be32 daddr, - __be16 sport, __be16 dport); -u32 secure_tcp_ts_off(const struct net *net, __be32 saddr, __be32 daddr); -u32 secure_tcpv6_seq(const __be32 *saddr, const __be32 *daddr, - __be16 sport, __be16 dport); -u32 secure_tcpv6_ts_off(const struct net *net, - const __be32 *saddr, const __be32 *daddr); u64 secure_dccp_sequence_number(__be32 saddr, __be32 daddr, __be16 sport, __be16 dport); u64 secure_dccpv6_sequence_number(__be32 *saddr, __be32 *daddr, __be16 sport, __be16 dport); +union tcp_seq_and_ts_off +secure_tcp_seq_and_ts_off(const struct net *net, __be32 saddr, __be32 daddr, + __be16 sport, __be16 dport); + +static inline u32 secure_tcp_seq(__be32 saddr, __be32 daddr, + __be16 sport, __be16 dport) +{ + union tcp_seq_and_ts_off ts; + + ts = secure_tcp_seq_and_ts_off(&init_net, saddr, daddr, + sport, dport); + + return ts.seq; +} + +union tcp_seq_and_ts_off +secure_tcpv6_seq_and_ts_off(const struct net *net, const __be32 *saddr, + const __be32 *daddr, + __be16 sport, __be16 dport); + +static inline u32 secure_tcpv6_seq(const __be32 *saddr, const __be32 *daddr, + __be16 sport, __be16 dport) +{ + union tcp_seq_and_ts_off ts; + + ts = secure_tcpv6_seq_and_ts_off(&init_net, saddr, daddr, + sport, dport); + return ts.seq; +} #endif /* _NET_SECURE_SEQ */ diff --git a/include/net/tcp.h b/include/net/tcp.h index e8f81924defc..f0a6468a1c34 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -44,6 +44,7 @@ #ifndef __GENKSYMS__ #include <net/xfrm.h> #endif +#include <net/secure_seq.h> #include <linux/seq_file.h> #include <linux/memcontrol.h> @@ -2180,8 +2181,9 @@ struct tcp_request_sock_ops { struct sk_buff *skb, struct flowi *fl, struct request_sock *req); - u32 (*init_seq)(const struct sk_buff *skb); - u32 (*init_ts_off)(const struct net *net, const struct sk_buff *skb); + union tcp_seq_and_ts_off (*init_seq_and_ts_off)( + const struct net *net, + const struct sk_buff *skb); int (*send_synack)(const struct sock *sk, struct dst_entry *dst, struct flowi *fl, struct request_sock *req, struct tcp_fastopen_cookie *foc, diff --git a/net/core/secure_seq.c b/net/core/secure_seq.c index b0ff6153be62..740642aeaf76 100644 --- a/net/core/secure_seq.c +++ b/net/core/secure_seq.c @@ -20,7 +20,6 @@ #include <net/tcp.h> static siphash_aligned_key_t net_secret; -static siphash_aligned_key_t ts_secret; #define EPHEMERAL_PORT_SHUFFLE_PERIOD (10 * HZ) @@ -28,11 +27,6 @@ static __always_inline void net_secret_init(void) { net_get_random_once(&net_secret, sizeof(net_secret)); } - -static __always_inline void ts_secret_init(void) -{ - net_get_random_once(&ts_secret, sizeof(ts_secret)); -} #endif #ifdef CONFIG_INET @@ -53,28 +47,9 @@ static u32 seq_scale(u32 seq) #endif #if IS_ENABLED(CONFIG_IPV6) -u32 secure_tcpv6_ts_off(const struct net *net, - const __be32 *saddr, const __be32 *daddr) -{ - const struct { - struct in6_addr saddr; - struct in6_addr daddr; - } __aligned(SIPHASH_ALIGNMENT) combined = { - .saddr = *(struct in6_addr *)saddr, - .daddr = *(struct in6_addr *)daddr, - }; - - if (READ_ONCE(net->ipv4.sysctl_tcp_timestamps) != 1) - return 0; - - ts_secret_init(); - return siphash(&combined, offsetofend(typeof(combined), daddr), - &ts_secret); -} -EXPORT_SYMBOL(secure_tcpv6_ts_off); - -u32 secure_tcpv6_seq(const __be32 *saddr, const __be32 *daddr, - __be16 sport, __be16 dport) +union tcp_seq_and_ts_off +secure_tcpv6_seq_and_ts_off(const struct net *net, const __be32 *saddr, + const __be32 *daddr, __be16 sport, __be16 dport) { const struct { struct in6_addr saddr; @@ -87,14 +62,20 @@ u32 secure_tcpv6_seq(const __be32 *saddr, const __be32 *daddr, .sport = sport, .dport = dport }; - u32 hash; + union tcp_seq_and_ts_off st; net_secret_init(); - hash = siphash(&combined, offsetofend(typeof(combined), dport), - &net_secret); - return seq_scale(hash); + + st.hash64 = siphash(&combined, offsetofend(typeof(combined), dport), + &net_secret); + + if (READ_ONCE(net->ipv4.sysctl_tcp_timestamps) != 1) + st.ts_off = 0; + + st.seq = seq_scale(st.seq); + return st; } -EXPORT_SYMBOL(secure_tcpv6_seq); +EXPORT_SYMBOL(secure_tcpv6_seq_and_ts_off); u64 secure_ipv6_port_ephemeral(const __be32 *saddr, const __be32 *daddr, __be16 dport) @@ -118,33 +99,30 @@ EXPORT_SYMBOL(secure_ipv6_port_ephemeral); #endif #ifdef CONFIG_INET -u32 secure_tcp_ts_off(const struct net *net, __be32 saddr, __be32 daddr) -{ - if (READ_ONCE(net->ipv4.sysctl_tcp_timestamps) != 1) - return 0; - - ts_secret_init(); - return siphash_2u32((__force u32)saddr, (__force u32)daddr, - &ts_secret); -} - /* secure_tcp_seq_and_tsoff(a, b, 0, d) == secure_ipv4_port_ephemeral(a, b, d), * but fortunately, `sport' cannot be 0 in any circumstances. If this changes, * it would be easy enough to have the former function use siphash_4u32, passing * the arguments as separate u32. */ -u32 secure_tcp_seq(__be32 saddr, __be32 daddr, - __be16 sport, __be16 dport) +union tcp_seq_and_ts_off +secure_tcp_seq_and_ts_off(const struct net *net, __be32 saddr, __be32 daddr, + __be16 sport, __be16 dport) { - u32 hash; + u32 ports = (__force u32)sport << 16 | (__force u32)dport; + union tcp_seq_and_ts_off st; net_secret_init(); - hash = siphash_3u32((__force u32)saddr, (__force u32)daddr, - (__force u32)sport << 16 | (__force u32)dport, - &net_secret); - return seq_scale(hash); + + st.hash64 = siphash_3u32((__force u32)saddr, (__force u32)daddr, + ports, &net_secret); + + if (READ_ONCE(net->ipv4.sysctl_tcp_timestamps) != 1) + st.ts_off = 0; + + st.seq = seq_scale(st.seq); + return st; } -EXPORT_SYMBOL_GPL(secure_tcp_seq); +EXPORT_SYMBOL_GPL(secure_tcp_seq_and_ts_off); u64 secure_ipv4_port_ephemeral(__be32 saddr, __be32 daddr, __be16 dport) { diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c index 64f098d127ac..b8f7e2b0b4f3 100644 --- a/net/ipv4/syncookies.c +++ b/net/ipv4/syncookies.c @@ -353,9 +353,14 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb) tcp_parse_options(sock_net(sk), skb, &tcp_opt, 0, NULL); if (tcp_opt.saw_tstamp && tcp_opt.rcv_tsecr) { - tsoff = secure_tcp_ts_off(sock_net(sk), - ip_hdr(skb)->daddr, - ip_hdr(skb)->saddr); + union tcp_seq_and_ts_off st; + + st = secure_tcp_seq_and_ts_off(sock_net(sk), + ip_hdr(skb)->daddr, + ip_hdr(skb)->saddr, + tcp_hdr(skb)->dest, + tcp_hdr(skb)->source); + tsoff = st.ts_off; tcp_opt.rcv_tsecr -= tsoff; } diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 8d6757ec06aa..9ca0a78d5a4b 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -7165,6 +7165,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, struct tcp_sock *tp = tcp_sk(sk); struct net *net = sock_net(sk); struct sock *fastopen_sk = NULL; + union tcp_seq_and_ts_off st; struct request_sock *req; bool want_cookie = false; struct dst_entry *dst; @@ -7227,9 +7228,12 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, if (!dst) goto drop_and_free; - if (tmp_opt.tstamp_ok) - tcp_rsk(req)->ts_off = af_ops->init_ts_off(net, skb); + if (tmp_opt.tstamp_ok || (!want_cookie && !isn)) + st = af_ops->init_seq_and_ts_off(net, skb); + if (tmp_opt.tstamp_ok) { + tcp_rsk(req)->ts_off = st.ts_off; + } if (!want_cookie && !isn) { int max_syn_backlog = READ_ONCE(net->ipv4.sysctl_max_syn_backlog); @@ -7250,7 +7254,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, goto drop_and_release; } - isn = af_ops->init_seq(skb); + isn = st.seq; } tcp_ecn_create_request(req, skb, sk, dst); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 56d00db07d12..ceb34b30218b 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -96,17 +96,14 @@ static DEFINE_PER_CPU(struct sock *, ipv4_tcp_sk); static DEFINE_MUTEX(tcp_exit_batch_mutex); -static u32 tcp_v4_init_seq(const struct sk_buff *skb) +static union tcp_seq_and_ts_off +tcp_v4_init_seq_and_ts_off(const struct net *net, const struct sk_buff *skb) { - return secure_tcp_seq(ip_hdr(skb)->daddr, - ip_hdr(skb)->saddr, - tcp_hdr(skb)->dest, - tcp_hdr(skb)->source); -} - -static u32 tcp_v4_init_ts_off(const struct net *net, const struct sk_buff *skb) -{ - return secure_tcp_ts_off(net, ip_hdr(skb)->daddr, ip_hdr(skb)->saddr); + return secure_tcp_seq_and_ts_off(net, + ip_hdr(skb)->daddr, + ip_hdr(skb)->saddr, + tcp_hdr(skb)->dest, + tcp_hdr(skb)->source); } int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp) @@ -313,15 +310,16 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len) rt = NULL; if (likely(!tp->repair)) { + union tcp_seq_and_ts_off st; + + st = secure_tcp_seq_and_ts_off(net, + inet->inet_saddr, + inet->inet_daddr, + inet->inet_sport, + usin->sin_port); if (!tp->write_seq) - WRITE_ONCE(tp->write_seq, - secure_tcp_seq(inet->inet_saddr, - inet->inet_daddr, - inet->inet_sport, - usin->sin_port)); - WRITE_ONCE(tp->tsoffset, - secure_tcp_ts_off(net, inet->inet_saddr, - inet->inet_daddr)); + WRITE_ONCE(tp->write_seq, st.seq); + WRITE_ONCE(tp->tsoffset, st.ts_off); } atomic_set(&inet->inet_id, get_random_u16()); @@ -1535,8 +1533,7 @@ const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = { .cookie_init_seq = cookie_v4_init_sequence, #endif .route_req = tcp_v4_route_req, - .init_seq = tcp_v4_init_seq, - .init_ts_off = tcp_v4_init_ts_off, + .init_seq_and_ts_off = tcp_v4_init_seq_and_ts_off, .send_synack = tcp_v4_send_synack, }; diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c index a6cefee15005..bdbd3fd20448 100644 --- a/net/ipv6/syncookies.c +++ b/net/ipv6/syncookies.c @@ -161,9 +161,14 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb) tcp_parse_options(sock_net(sk), skb, &tcp_opt, 0, NULL); if (tcp_opt.saw_tstamp && tcp_opt.rcv_tsecr) { - tsoff = secure_tcpv6_ts_off(sock_net(sk), - ipv6_hdr(skb)->daddr.s6_addr32, - ipv6_hdr(skb)->saddr.s6_addr32); + union tcp_seq_and_ts_off st; + + st = secure_tcpv6_seq_and_ts_off(sock_net(sk), + ipv6_hdr(skb)->daddr.s6_addr32, + ipv6_hdr(skb)->saddr.s6_addr32, + tcp_hdr(skb)->dest, + tcp_hdr(skb)->source); + tsoff = st.ts_off; tcp_opt.rcv_tsecr -= tsoff; } diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index f285e52b8b85..793e644fa750 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -107,18 +107,14 @@ static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb) } } -static u32 tcp_v6_init_seq(const struct sk_buff *skb) +static union tcp_seq_and_ts_off +tcp_v6_init_seq_and_ts_off(const struct net *net, const struct sk_buff *skb) { - return secure_tcpv6_seq(ipv6_hdr(skb)->daddr.s6_addr32, - ipv6_hdr(skb)->saddr.s6_addr32, - tcp_hdr(skb)->dest, - tcp_hdr(skb)->source); -} - -static u32 tcp_v6_init_ts_off(const struct net *net, const struct sk_buff *skb) -{ - return secure_tcpv6_ts_off(net, ipv6_hdr(skb)->daddr.s6_addr32, - ipv6_hdr(skb)->saddr.s6_addr32); + return secure_tcpv6_seq_and_ts_off(net, + ipv6_hdr(skb)->daddr.s6_addr32, + ipv6_hdr(skb)->saddr.s6_addr32, + tcp_hdr(skb)->dest, + tcp_hdr(skb)->source); } static int tcp_v6_pre_connect(struct sock *sk, struct sockaddr *uaddr, @@ -318,14 +314,16 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr, sk_set_txhash(sk); if (likely(!tp->repair)) { + union tcp_seq_and_ts_off st; + + st = secure_tcpv6_seq_and_ts_off(net, + np->saddr.s6_addr32, + sk->sk_v6_daddr.s6_addr32, + inet->inet_sport, + inet->inet_dport); if (!tp->write_seq) - WRITE_ONCE(tp->write_seq, - secure_tcpv6_seq(np->saddr.s6_addr32, - sk->sk_v6_daddr.s6_addr32, - inet->inet_sport, - inet->inet_dport)); - tp->tsoffset = secure_tcpv6_ts_off(net, np->saddr.s6_addr32, - sk->sk_v6_daddr.s6_addr32); + WRITE_ONCE(tp->write_seq, st.seq); + tp->tsoffset = st.ts_off; } if (tcp_fastopen_defer_connect(sk, &err)) @@ -831,8 +829,7 @@ const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = { .cookie_init_seq = cookie_v6_init_sequence, #endif .route_req = tcp_v6_route_req, - .init_seq = tcp_v6_init_seq, - .init_ts_off = tcp_v6_init_ts_off, + .init_seq_and_ts_off = tcp_v6_init_seq_and_ts_off, .send_synack = tcp_v6_send_synack, }; -- 2.34.1

2 1

[PATCH OLK-6.6] net: gro: fix outer network offset
by Li Xiasong 02 Apr '26

02 Apr '26

From: Paolo Abeni <pabeni(a)redhat.com> stable inclusion from stable-v6.6.124 commit 9d40a85138568696387ef04cd004c64612a70874 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/13895 CVE: CVE-2026-23254 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 5c2c3c38be396257a6a2e55bd601a12bb9781507 ] The udp GRO complete stage assumes that all the packets inserted the RX have the `encapsulation` flag zeroed. Such assumption is not true, as a few H/W NICs can set such flag when H/W offloading the checksum for an UDP encapsulated traffic, the tun driver can inject GSO packets with UDP encapsulation and the problematic layout can also be created via a veth based setup. Due to the above, in the problematic scenarios, udp4_gro_complete() uses the wrong network offset (inner instead of outer) to compute the outer UDP header pseudo checksum, leading to csum validation errors later on in packet processing. Address the issue always clearing the encapsulation flag at GRO completion time. Such flag will be set again as needed for encapsulated packets by udp_gro_complete(). Fixes: 5ef31ea5d053 ("net: gro: fix udp bad offset in socket lookup by adding {inner_}network_offset to napi_gro_cb") Reviewed-by: Willem de Bruijn <willemb(a)google.com> Signed-off-by: Paolo Abeni <pabeni(a)redhat.com> Reviewed-by: Eric Dumazet <edumazet(a)google.com> Link: https://patch.msgid.link/562638dbebb3b15424220e26a180274b387e2a88.177003208… Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Li Xiasong <lixiasong1(a)huawei.com> --- net/core/gro.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/net/core/gro.c b/net/core/gro.c index 2f72cce0b219..98f47ff5837b 100644 --- a/net/core/gro.c +++ b/net/core/gro.c @@ -242,6 +242,8 @@ static void napi_gro_complete(struct napi_struct *napi, struct sk_buff *skb) goto out; } + /* NICs can feed encapsulated packets into GRO */ + skb->encapsulation = 0; rcu_read_lock(); list_for_each_entry_rcu(ptype, head, list) { if (ptype->type != type || !ptype->callbacks.gro_complete) -- 2.34.1

2 1

[PATCH OLK-6.6] net: annotate data-races around sk->sk_{data_ready,write_space}
by Li Xiasong 02 Apr '26

02 Apr '26

From: Eric Dumazet <edumazet(a)google.com> mainline inclusion from mainline-v7.0-rc3 commit 2ef2b20cf4e04ac8a6ba68493f8780776ff84300 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/13979 CVE: CVE-2026-23302 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- skmsg (and probably other layers) are changing these pointers while other cpus might read them concurrently. Add corresponding READ_ONCE()/WRITE_ONCE() annotations for UDP, TCP and AF_UNIX. Fixes: 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface") Reported-by: syzbot+87f770387a9e5dc6b79b(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/699ee9fc.050a0220.1cd54b.0009.GAE@google.com/ Signed-off-by: Eric Dumazet <edumazet(a)google.com> Cc: Daniel Borkmann <daniel(a)iogearbox.net> Cc: John Fastabend <john.fastabend(a)gmail.com> Cc: Jakub Sitnicki <jakub(a)cloudflare.com> Cc: Willem de Bruijn <willemdebruijn.kernel(a)gmail.com> Reviewed-by: Kuniyuki Iwashima <kuniyu(a)google.com> Link: https://patch.msgid.link/20260225131547.1085509-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Conflicts: net/ipv4/udp.c net/ipv4/tcp.c net/ipv4/tcp_input.c net/ipv4/tcp_minisocks.c net/unix/af_unix.c [conflicts due to not merge bd61848900bf ("net: devmem: Implement TX path") and 95b9a87c6a6b ("tcp: record last received ipv6 flowlabel") and f0db2bca0cf9 ("tcp: rework {__,}tcp_ecn_check_ce() -> tcp_data_ecn_check()") and 2bd99aef1b19 ("tcp: accept bare FIN packets under memory pressure") and b98256959305 ("tcp: make the dropreason really work when calling tcp_rcv_state_process()") and 85cb0757d7e1 ("net: Convert proto_ops connect() callbacks to use sockaddr_unsized") and f4e1fb04c123 ("af_unix: Use cached value for SOCK_STREAM in unix_inq_len().") and b650bf0977d3 ("udp: remove busylock and add per NUMA queues").] Signed-off-by: Li Xiasong <lixiasong1(a)huawei.com> --- net/core/skmsg.c | 14 +++++++------- net/ipv4/tcp.c | 4 ++-- net/ipv4/tcp_bpf.c | 2 +- net/ipv4/tcp_input.c | 14 ++++++++------ net/ipv4/tcp_minisocks.c | 2 +- net/ipv4/udp.c | 3 ++- net/ipv4/udp_bpf.c | 2 +- net/unix/af_unix.c | 8 ++++---- 8 files changed, 26 insertions(+), 23 deletions(-) diff --git a/net/core/skmsg.c b/net/core/skmsg.c index 3429c6e2ce21..e3696d510eec 100644 --- a/net/core/skmsg.c +++ b/net/core/skmsg.c @@ -1174,8 +1174,8 @@ void sk_psock_start_strp(struct sock *sk, struct sk_psock *psock) return; psock->saved_data_ready = sk->sk_data_ready; - sk->sk_data_ready = sk_psock_strp_data_ready; - sk->sk_write_space = sk_psock_write_space; + WRITE_ONCE(sk->sk_data_ready, sk_psock_strp_data_ready); + WRITE_ONCE(sk->sk_write_space, sk_psock_write_space); } void sk_psock_stop_strp(struct sock *sk, struct sk_psock *psock) @@ -1185,8 +1185,8 @@ void sk_psock_stop_strp(struct sock *sk, struct sk_psock *psock) if (!psock->saved_data_ready) return; - sk->sk_data_ready = psock->saved_data_ready; - psock->saved_data_ready = NULL; + WRITE_ONCE(sk->sk_data_ready, psock->saved_data_ready); + WRITE_ONCE(psock->saved_data_ready, NULL); strp_stop(&psock->strp); } @@ -1265,8 +1265,8 @@ void sk_psock_start_verdict(struct sock *sk, struct sk_psock *psock) return; psock->saved_data_ready = sk->sk_data_ready; - sk->sk_data_ready = sk_psock_verdict_data_ready; - sk->sk_write_space = sk_psock_write_space; + WRITE_ONCE(sk->sk_data_ready, sk_psock_verdict_data_ready); + WRITE_ONCE(sk->sk_write_space, sk_psock_write_space); } void sk_psock_stop_verdict(struct sock *sk, struct sk_psock *psock) @@ -1277,6 +1277,6 @@ void sk_psock_stop_verdict(struct sock *sk, struct sk_psock *psock) if (!psock->saved_data_ready) return; - sk->sk_data_ready = psock->saved_data_ready; + WRITE_ONCE(sk->sk_data_ready, psock->saved_data_ready); psock->saved_data_ready = NULL; } diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index bf23aa827031..1fe89e742f75 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -1342,7 +1342,7 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size) err = sk_stream_error(sk, flags, err); /* make sure we wake any epoll edge trigger waiter */ if (unlikely(tcp_rtx_and_write_queues_empty(sk) && err == -EAGAIN)) { - sk->sk_write_space(sk); + READ_ONCE(sk->sk_write_space)(sk); tcp_chrono_stop(sk, TCP_CHRONO_SNDBUF_LIMITED); } return err; @@ -3705,7 +3705,7 @@ int do_tcp_setsockopt(struct sock *sk, int level, int optname, break; case TCP_NOTSENT_LOWAT: WRITE_ONCE(tp->notsent_lowat, val); - sk->sk_write_space(sk); + READ_ONCE(sk->sk_write_space)(sk); break; case TCP_INQ: if (val > 1 || val < 0) diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c index 0e56f3baf98a..9ceb6e0c8b00 100644 --- a/net/ipv4/tcp_bpf.c +++ b/net/ipv4/tcp_bpf.c @@ -668,7 +668,7 @@ int tcp_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore) WRITE_ONCE(sk->sk_prot->unhash, psock->saved_unhash); tcp_update_ulp(sk, psock->sk_proto, psock->saved_write_space); } else { - sk->sk_write_space = psock->saved_write_space; + WRITE_ONCE(sk->sk_write_space, psock->saved_write_space); /* Pairs with lockless read in sk_clone_lock() */ sock_replace_proto(sk, psock->sk_proto); } diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 8d6757ec06aa..a10e7da1a612 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4985,7 +4985,7 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb) if (unlikely(tcp_try_rmem_schedule(sk, skb, skb->truesize))) { NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPOFODROP); - sk->sk_data_ready(sk); + READ_ONCE(sk->sk_data_ready)(sk); tcp_drop_reason(sk, skb, SKB_DROP_REASON_PROTO_MEM); return; } @@ -5192,7 +5192,7 @@ int tcp_send_rcvq(struct sock *sk, struct msghdr *msg, size_t size) void tcp_data_ready(struct sock *sk) { if (tcp_epollin_ready(sk, sk->sk_rcvlowat) || sock_flag(sk, SOCK_DONE)) - sk->sk_data_ready(sk); + READ_ONCE(sk->sk_data_ready)(sk); } static void tcp_data_queue(struct sock *sk, struct sk_buff *skb) @@ -5238,7 +5238,7 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb) inet_csk(sk)->icsk_ack.pending |= (ICSK_ACK_NOMEM | ICSK_ACK_NOW); inet_csk_schedule_ack(sk); - sk->sk_data_ready(sk); + READ_ONCE(sk->sk_data_ready)(sk); if (skb_queue_len(&sk->sk_receive_queue)) { reason = SKB_DROP_REASON_PROTO_MEM; @@ -5675,7 +5675,9 @@ static void tcp_new_space(struct sock *sk) tp->snd_cwnd_stamp = tcp_jiffies32; } - INDIRECT_CALL_1(sk->sk_write_space, sk_stream_write_space, sk); + INDIRECT_CALL_1(READ_ONCE(sk->sk_write_space), + sk_stream_write_space, + sk); } /* Caller made space either from: @@ -5873,7 +5875,7 @@ static void tcp_urg(struct sock *sk, struct sk_buff *skb, const struct tcphdr *t BUG(); WRITE_ONCE(tp->urg_data, TCP_URG_VALID | tmp); if (!sock_flag(sk, SOCK_DEAD)) - sk->sk_data_ready(sk); + READ_ONCE(sk->sk_data_ready)(sk); } } } @@ -7279,7 +7281,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, sock_put(fastopen_sk); goto drop_and_free; } - sk->sk_data_ready(sk); + READ_ONCE(sk->sk_data_ready)(sk); bh_unlock_sock(fastopen_sk); sock_put(fastopen_sk); } else { diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 770b1ffb2e22..8d72532143bf 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -900,7 +900,7 @@ int tcp_child_process(struct sock *parent, struct sock *child, ret = tcp_rcv_state_process(child, skb); /* Wakeup parent, send SIGIO */ if (state == TCP_SYN_RECV && child->sk_state != state) - parent->sk_data_ready(parent); + READ_ONCE(parent->sk_data_ready)(parent); } else { /* Alas, it is possible again, because we do lookup * in main socket hash table and lock on listening diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 4c1f82f6777f..40100fed5c67 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1618,7 +1618,8 @@ int __udp_enqueue_schedule_skb(struct sock *sk, struct sk_buff *skb) spin_unlock(&list->lock); if (!sock_flag(sk, SOCK_DEAD)) - INDIRECT_CALL_1(sk->sk_data_ready, sock_def_readable, sk); + INDIRECT_CALL_1(READ_ONCE(sk->sk_data_ready), + sock_def_readable, sk); busylock_release(busy); return 0; diff --git a/net/ipv4/udp_bpf.c b/net/ipv4/udp_bpf.c index 0735d820e413..44271ba1adec 100644 --- a/net/ipv4/udp_bpf.c +++ b/net/ipv4/udp_bpf.c @@ -143,7 +143,7 @@ int udp_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore) int family = sk->sk_family == AF_INET ? UDP_BPF_IPV4 : UDP_BPF_IPV6; if (restore) { - sk->sk_write_space = psock->saved_write_space; + WRITE_ONCE(sk->sk_write_space, psock->saved_write_space); sock_replace_proto(sk, psock->sk_proto); return 0; } diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c index abf9fe35e620..51f444ecebf6 100644 --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -1678,7 +1678,7 @@ static int unix_stream_connect(struct socket *sock, struct sockaddr *uaddr, __skb_queue_tail(&other->sk_receive_queue, skb); spin_unlock(&other->sk_receive_queue.lock); unix_state_unlock(other); - other->sk_data_ready(other); + READ_ONCE(other->sk_data_ready)(other); sock_put(other); return 0; @@ -2138,7 +2138,7 @@ static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg, scm_stat_add(other, skb); skb_queue_tail(&other->sk_receive_queue, skb); unix_state_unlock(other); - other->sk_data_ready(other); + READ_ONCE(other->sk_data_ready)(other); sock_put(other); scm_destroy(&scm); return len; @@ -2206,7 +2206,7 @@ static int queue_oob(struct socket *sock, struct msghdr *msg, struct sock *other sk_send_sigurg(other); unix_state_unlock(other); - other->sk_data_ready(other); + READ_ONCE(other->sk_data_ready)(other); return err; } @@ -2317,7 +2317,7 @@ static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg, scm_stat_add(other, skb); skb_queue_tail(&other->sk_receive_queue, skb); unix_state_unlock(other); - other->sk_data_ready(other); + READ_ONCE(other->sk_data_ready)(other); sent += size; } -- 2.34.1

2 1

[PATCH OLK-6.6] net: add proper RCU protection to /proc/net/ptype
by Li Xiasong 02 Apr '26

02 Apr '26

From: Eric Dumazet <edumazet(a)google.com> mainline inclusion from mainline-v6.19 commit f613e8b4afea0cd17c7168e8b00e25bc8d33175d category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/13893 CVE: CVE-2026-23255 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- Yin Fengwei reported an RCU stall in ptype_seq_show() and provided a patch. Real issue is that ptype_seq_next() and ptype_seq_show() violate RCU rules. ptype_seq_show() runs under rcu_read_lock(), and reads pt->dev to get device name without any barrier. At the same time, concurrent writers can remove a packet_type structure (which is correctly freed after an RCU grace period) and clear pt->dev without an RCU grace period. Define ptype_iter_state to carry a dev pointer along seq_net_private: struct ptype_iter_state { struct seq_net_private p; struct net_device *dev; // added in this patch }; We need to record the device pointer in ptype_get_idx() and ptype_seq_next() so that ptype_seq_show() is safe against concurrent pt->dev changes. We also need to add full RCU protection in ptype_seq_next(). (Missing READ_ONCE() when reading list.next values) Many thanks to Dong Chenchen for providing a repro. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: 1d10f8a1f40b ("net-procfs: show net devices bound packet types") Fixes: c353e8983e0d ("net: introduce per netns packet chains") Reported-by: Yin Fengwei <fengwei_yin(a)linux.alibaba.com> Reported-by: Dong Chenchen <dongchenchen2(a)huawei.com> Closes: https://lore.kernel.org/netdev/CANn89iKRRKPnWjJmb-_3a=sq+9h6DvTQM4DBZHT5ZRG… Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: Willem de Bruijn <willemb(a)google.com> Tested-by: Yin Fengwei <fengwei_yin(a)linux.alibaba.com> Link: https://patch.msgid.link/20260202205217.2881198-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Conflicts: net/core/net-procfs.c [conflicts due to not merge c353e8983e0d ("net: introduce per netns packet chains").] Signed-off-by: Li Xiasong <lixiasong1(a)huawei.com> --- net/core/net-procfs.c | 48 +++++++++++++++++++++++++++++-------------- 1 file changed, 33 insertions(+), 15 deletions(-) diff --git a/net/core/net-procfs.c b/net/core/net-procfs.c index 09f7ed1a04e8..caad20464a4d 100644 --- a/net/core/net-procfs.c +++ b/net/core/net-procfs.c @@ -200,8 +200,14 @@ static const struct seq_operations softnet_seq_ops = { .show = softnet_seq_show, }; +struct ptype_iter_state { + struct seq_net_private p; + struct net_device *dev; +}; + static void *ptype_get_idx(struct seq_file *seq, loff_t pos) { + struct ptype_iter_state *iter = seq->private; struct list_head *ptype_list = NULL; struct packet_type *pt = NULL; struct net_device *dev; @@ -211,12 +217,16 @@ static void *ptype_get_idx(struct seq_file *seq, loff_t pos) for_each_netdev_rcu(seq_file_net(seq), dev) { ptype_list = &dev->ptype_all; list_for_each_entry_rcu(pt, ptype_list, list) { - if (i == pos) + if (i == pos) { + iter->dev = dev; return pt; + } ++i; } } + iter->dev = NULL; + list_for_each_entry_rcu(pt, &ptype_all, list) { if (i == pos) return pt; @@ -242,6 +252,7 @@ static void *ptype_seq_start(struct seq_file *seq, loff_t *pos) static void *ptype_seq_next(struct seq_file *seq, void *v, loff_t *pos) { + struct ptype_iter_state *iter = seq->private; struct net_device *dev; struct packet_type *pt; struct list_head *nxt; @@ -252,20 +263,22 @@ static void *ptype_seq_next(struct seq_file *seq, void *v, loff_t *pos) return ptype_get_idx(seq, 0); pt = v; - nxt = pt->list.next; - if (pt->dev) { - if (nxt != &pt->dev->ptype_all) + nxt = READ_ONCE(pt->list.next); + dev = iter->dev; + if (dev) { + if (nxt != &dev->ptype_all) goto found; - dev = pt->dev; for_each_netdev_continue_rcu(seq_file_net(seq), dev) { - if (!list_empty(&dev->ptype_all)) { - nxt = dev->ptype_all.next; + nxt = READ_ONCE(dev->ptype_all.next); + if (nxt != &dev->ptype_all) { + iter->dev = dev; goto found; } } - nxt = ptype_all.next; + iter->dev = NULL; + nxt = READ_ONCE(ptype_all.next); goto ptype_all; } @@ -274,14 +287,14 @@ static void *ptype_seq_next(struct seq_file *seq, void *v, loff_t *pos) if (nxt != &ptype_all) goto found; hash = 0; - nxt = ptype_base[0].next; + nxt = READ_ONCE(ptype_base[0].next); } else hash = ntohs(pt->type) & PTYPE_HASH_MASK; while (nxt == &ptype_base[hash]) { if (++hash >= PTYPE_HASH_SIZE) return NULL; - nxt = ptype_base[hash].next; + nxt = READ_ONCE(ptype_base[hash].next); } found: return list_entry(nxt, struct packet_type, list); @@ -295,19 +308,24 @@ static void ptype_seq_stop(struct seq_file *seq, void *v) static int ptype_seq_show(struct seq_file *seq, void *v) { + struct ptype_iter_state *iter = seq->private; struct packet_type *pt = v; + struct net_device *dev; - if (v == SEQ_START_TOKEN) + if (v == SEQ_START_TOKEN) { seq_puts(seq, "Type Device Function\n"); - else if ((!pt->af_packet_net || net_eq(pt->af_packet_net, seq_file_net(seq))) && - (!pt->dev || net_eq(dev_net(pt->dev), seq_file_net(seq)))) { + return 0; + } + dev = iter->dev; + if ((!pt->af_packet_net || net_eq(pt->af_packet_net, seq_file_net(seq))) && + (!dev || net_eq(dev_net(dev), seq_file_net(seq)))) { if (pt->type == htons(ETH_P_ALL)) seq_puts(seq, "ALL "); else seq_printf(seq, "%04x", ntohs(pt->type)); seq_printf(seq, " %-8s %ps\n", - pt->dev ? pt->dev->name : "", pt->func); + dev ? dev->name : "", pt->func); } return 0; @@ -331,7 +349,7 @@ static int __net_init dev_proc_net_init(struct net *net) &softnet_seq_ops)) goto out_dev; if (!proc_create_net("ptype", 0444, net->proc_net, &ptype_seq_ops, - sizeof(struct seq_net_private))) + sizeof(struct ptype_iter_state))) goto out_softnet; if (wext_proc_init(net)) -- 2.34.1

2 1

[PATCH OLK-6.6] af_unix: Give up GC if MSG_PEEK intervened.
by Li Xiasong 02 Apr '26

02 Apr '26

From: Kuniyuki Iwashima <kuniyu(a)google.com> mainline inclusion from mainline-v7.0-rc5 commit e5b31d988a41549037b8d8721a3c3cae893d8670 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/14025 CVE: CVE-2026-23394 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- Igor Ushakov reported that GC purged the receive queue of an alive socket due to a race with MSG_PEEK with a nice repro. This is the exact same issue previously fixed by commit cbcf01128d0a ("af_unix: fix garbage collect vs MSG_PEEK"). After GC was replaced with the current algorithm, the cited commit removed the locking dance in unix_peek_fds() and reintroduced the same issue. The problem is that MSG_PEEK bumps a file refcount without interacting with GC. Consider an SCC containing sk-A and sk-B, where sk-A is close()d but can be recv()ed via sk-B. The bad thing happens if sk-A is recv()ed with MSG_PEEK from sk-B and sk-B is close()d while GC is checking unix_vertex_dead() for sk-A and sk-B. GC thread User thread --------- ----------- unix_vertex_dead(sk-A) -> true <------. \ `------ recv(sk-B, MSG_PEEK) invalidate !! -> sk-A's file refcount : 1 -> 2 close(sk-B) -> sk-B's file refcount : 2 -> 1 unix_vertex_dead(sk-B) -> true Initially, sk-A's file refcount is 1 by the inflight fd in sk-B recvq. GC thinks sk-A is dead because the file refcount is the same as the number of its inflight fds. However, sk-A's file refcount is bumped silently by MSG_PEEK, which invalidates the previous evaluation. At this moment, sk-B's file refcount is 2; one by the open fd, and one by the inflight fd in sk-A. The subsequent close() releases one refcount by the former. Finally, GC incorrectly concludes that both sk-A and sk-B are dead. One option is to restore the locking dance in unix_peek_fds(), but we can resolve this more elegantly thanks to the new algorithm. The point is that the issue does not occur without the subsequent close() and we actually do not need to synchronise MSG_PEEK with the dead SCC detection. When the issue occurs, close() and GC touch the same file refcount. If GC sees the refcount being decremented by close(), it can just give up garbage-collecting the SCC. Therefore, we only need to signal the race during MSG_PEEK with a proper memory barrier to make it visible to the GC. Let's use seqcount_t to notify GC when MSG_PEEK occurs and let it defer the SCC to the next run. This way no locking is needed on the MSG_PEEK side, and we can avoid imposing a penalty on every MSG_PEEK unnecessarily. Note that we can retry within unix_scc_dead() if MSG_PEEK is detected, but we do not do so to avoid hung task splat from abusive MSG_PEEK calls. Fixes: 118f457da9ed ("af_unix: Remove lock dance in unix_peek_fds().") Reported-by: Igor Ushakov <sysroot314(a)gmail.com> Signed-off-by: Kuniyuki Iwashima <kuniyu(a)google.com> Link: https://patch.msgid.link/20260311054043.1231316-1-kuniyu@google.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Conflicts: net/unix/af_unix.h include/net/af_unix.h net/unix/garbage.c [conflicts due to not merge 58b47c713711 ("af_unix: Count cyclic SCC.") and da8fc7a39be8 ("af_unix: Don't trigger GC from close() if unnecessary.") and 84960bf24031 ("af_unix: Move internal definitions to net/unix/.").] Signed-off-by: Li Xiasong <lixiasong1(a)huawei.com> --- include/net/af_unix.h | 1 + net/unix/af_unix.c | 2 ++ net/unix/garbage.c | 79 ++++++++++++++++++++++++++++--------------- 3 files changed, 54 insertions(+), 28 deletions(-) diff --git a/include/net/af_unix.h b/include/net/af_unix.h index b6eedf7650da..d2fa6d9f1e97 100644 --- a/include/net/af_unix.h +++ b/include/net/af_unix.h @@ -23,6 +23,7 @@ void unix_del_edges(struct scm_fp_list *fpl); void unix_update_edges(struct unix_sock *receiver); int unix_prepare_fpl(struct scm_fp_list *fpl); void unix_destroy_fpl(struct scm_fp_list *fpl); +void unix_peek_fpl(struct scm_fp_list *fpl); void unix_gc(void); void wait_for_unix_gc(struct scm_fp_list *fpl); diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c index abf9fe35e620..4407d4fedb02 100644 --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -1848,6 +1848,8 @@ static void unix_detach_fds(struct scm_cookie *scm, struct sk_buff *skb) static void unix_peek_fds(struct scm_cookie *scm, struct sk_buff *skb) { scm->fp = scm_fp_dup(UNIXCB(skb).fp); + + unix_peek_fpl(scm->fp); } static void unix_destruct_scm(struct sk_buff *skb) diff --git a/net/unix/garbage.c b/net/unix/garbage.c index 66fd606c43f4..1cdb54c61619 100644 --- a/net/unix/garbage.c +++ b/net/unix/garbage.c @@ -306,6 +306,25 @@ void unix_destroy_fpl(struct scm_fp_list *fpl) unix_free_vertices(fpl); } +static bool gc_in_progress; +static seqcount_t unix_peek_seq = SEQCNT_ZERO(unix_peek_seq); + +void unix_peek_fpl(struct scm_fp_list *fpl) +{ + static DEFINE_SPINLOCK(unix_peek_lock); + + if (!fpl || !fpl->count_unix) + return; + + if (!READ_ONCE(gc_in_progress)) + return; + + /* Invalidate the final refcnt check in unix_vertex_dead(). */ + spin_lock(&unix_peek_lock); + raw_write_seqcount_barrier(&unix_peek_seq); + spin_unlock(&unix_peek_lock); +} + static bool unix_vertex_dead(struct unix_vertex *vertex) { struct unix_edge *edge; @@ -339,6 +358,36 @@ static bool unix_vertex_dead(struct unix_vertex *vertex) return true; } +static LIST_HEAD(unix_visited_vertices); +static unsigned long unix_vertex_grouped_index = UNIX_VERTEX_INDEX_MARK2; + +static bool unix_scc_dead(struct list_head *scc, bool fast) +{ + struct unix_vertex *vertex; + bool scc_dead = true; + unsigned int seq; + + seq = read_seqcount_begin(&unix_peek_seq); + + list_for_each_entry_reverse(vertex, scc, scc_entry) { + /* Don't restart DFS from this vertex. */ + list_move_tail(&vertex->entry, &unix_visited_vertices); + + /* Mark vertex as off-stack for __unix_walk_scc(). */ + if (!fast) + vertex->index = unix_vertex_grouped_index; + + if (scc_dead) + scc_dead = unix_vertex_dead(vertex); + } + + /* If MSG_PEEK intervened, defer this SCC to the next round. */ + if (read_seqcount_retry(&unix_peek_seq, seq)) + return false; + + return scc_dead; +} + static void unix_collect_skb(struct list_head *scc, struct sk_buff_head *hitlist) { struct unix_vertex *vertex; @@ -392,9 +441,6 @@ static bool unix_scc_cyclic(struct list_head *scc) return false; } -static LIST_HEAD(unix_visited_vertices); -static unsigned long unix_vertex_grouped_index = UNIX_VERTEX_INDEX_MARK2; - static void __unix_walk_scc(struct unix_vertex *vertex, unsigned long *last_index, struct sk_buff_head *hitlist) { @@ -460,9 +506,7 @@ static void __unix_walk_scc(struct unix_vertex *vertex, unsigned long *last_inde } if (vertex->index == vertex->scc_index) { - struct unix_vertex *v; struct list_head scc; - bool scc_dead = true; /* SCC finalised. * @@ -471,18 +515,7 @@ static void __unix_walk_scc(struct unix_vertex *vertex, unsigned long *last_inde */ __list_cut_position(&scc, &vertex_stack, &vertex->scc_entry); - list_for_each_entry_reverse(v, &scc, scc_entry) { - /* Don't restart DFS from this vertex in unix_walk_scc(). */ - list_move_tail(&v->entry, &unix_visited_vertices); - - /* Mark vertex as off-stack. */ - v->index = unix_vertex_grouped_index; - - if (scc_dead) - scc_dead = unix_vertex_dead(v); - } - - if (scc_dead) { + if (unix_scc_dead(&scc, false)) { unix_collect_skb(&scc, hitlist); } else { if (unix_vertex_max_scc_index < vertex->scc_index) @@ -530,19 +563,11 @@ static void unix_walk_scc_fast(struct sk_buff_head *hitlist) while (!list_empty(&unix_unvisited_vertices)) { struct unix_vertex *vertex; struct list_head scc; - bool scc_dead = true; vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry); list_add(&scc, &vertex->scc_entry); - list_for_each_entry_reverse(vertex, &scc, scc_entry) { - list_move_tail(&vertex->entry, &unix_visited_vertices); - - if (scc_dead) - scc_dead = unix_vertex_dead(vertex); - } - - if (scc_dead) + if (unix_scc_dead(&scc, true)) unix_collect_skb(&scc, hitlist); else if (!unix_graph_maybe_cyclic) unix_graph_maybe_cyclic = unix_scc_cyclic(&scc); @@ -553,8 +578,6 @@ static void unix_walk_scc_fast(struct sk_buff_head *hitlist) list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices); } -static bool gc_in_progress; - static void __unix_gc(struct work_struct *work) { struct sk_buff_head hitlist; -- 2.34.1

2 1

[PATCH OLK-6.6 0/2] ip6_tunnel: fix skb_vlan_inet_prepare() return value handling regression
by Li Xiasong 02 Apr '26

02 Apr '26

This patchset contains a backport of the upstream change that introduced the issue, followed by the fix. Patch 1 is a backport of upstream commit f478b8239d65 ("net: tunnel: make skb_vlan_inet_prepare() return drop reasons") which changed the return value semantics of skb_vlan_inet_prepare(). Patch 2 adapts the return value handling in __ip6_tnl_rcv() to match the new semantics, fixing the regression. Including the upstream change as patch 1 ensures that applying both patches together does not introduce the issue that would occur if only patch 1 were merged. Li Xiasong (1): ip6_tunnel: adapt to skb_vlan_inet_prepare() return value change Menglong Dong (1): net: tunnel: make skb_vlan_inet_prepare() return drop reasons drivers/net/bareudp.c | 4 ++-- drivers/net/geneve.c | 4 ++-- include/net/ip_tunnels.h | 13 ++++++++----- net/ipv6/ip6_tunnel.c | 2 +- 4 files changed, 13 insertions(+), 10 deletions(-) -- 2.34.1

2 3

[PATCH OLK-5.10] KVM: x86/mmu: Drop/zap existing present SPTE even when creating an MMIO SPTE
by Zhang Kunbo 02 Apr '26

02 Apr '26

From: Sean Christopherson <seanjc(a)google.com> mainline inclusion from mainline-v7.0-rc6 commit aad885e774966e97b675dfe928da164214a71605 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/14045 CVE: CVE-2026-23401 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- When installing an emulated MMIO SPTE, do so *after* dropping/zapping the existing SPTE (if it's shadow-present). While commit a54aa15c6bda3 was right about it being impossible to convert a shadow-present SPTE to an MMIO SPTE due to a _guest_ write, it failed to account for writes to guest memory that are outside the scope of KVM. E.g. if host userspace modifies a shadowed gPTE to switch from a memslot to emulted MMIO and then the guest hits a relevant page fault, KVM will install the MMIO SPTE without first zapping the shadow-present SPTE. ------------[ cut here ]------------ is_shadow_present_pte(*sptep) WARNING: arch/x86/kvm/mmu/mmu.c:484 at mark_mmio_spte+0xb2/0xc0 [kvm], CPU#0: vmx_ept_stale_r/4292 Modules linked in: kvm_intel kvm irqbypass CPU: 0 UID: 1000 PID: 4292 Comm: vmx_ept_stale_r Not tainted 7.0.0-rc2-eafebd2d2ab0-sink-vm #319 PREEMPT Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 RIP: 0010:mark_mmio_spte+0xb2/0xc0 [kvm] Call Trace: <TASK> mmu_set_spte+0x237/0x440 [kvm] ept_page_fault+0x535/0x7f0 [kvm] kvm_mmu_do_page_fault+0xee/0x1f0 [kvm] kvm_mmu_page_fault+0x8d/0x620 [kvm] vmx_handle_exit+0x18c/0x5a0 [kvm_intel] kvm_arch_vcpu_ioctl_run+0xc55/0x1c20 [kvm] kvm_vcpu_ioctl+0x2d5/0x980 [kvm] __x64_sys_ioctl+0x8a/0xd0 do_syscall_64+0xb5/0x730 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x47fa3f </TASK> ---[ end trace 0000000000000000 ]--- Reported-by: Alexander Bulekov <bkov(a)amazon.com> Debugged-by: Alexander Bulekov <bkov(a)amazon.com> Suggested-by: Fred Griffoul <fgriffo(a)amazon.co.uk> Fixes: a54aa15c6bda3 ("KVM: x86/mmu: Handle MMIO SPTEs directly in mmu_set_spte()") Cc: stable(a)vger.kernel.org Signed-off-by: Sean Christopherson <seanjc(a)google.com> Conflicts: arch/x86/kvm/mmu/mmu.c [1075d41-not-merged,-and-context-conflicts] Signed-off-by: Zhang Kunbo <zhangkunbo(a)huawei.com> --- arch/x86/kvm/mmu/mmu.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index 0fee502a5f29..6ee54414295b 100755 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -2820,11 +2820,6 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, pgprintk("%s: spte %llx write_fault %d gfn %llx\n", __func__, *sptep, write_fault, gfn); - if (unlikely(is_noslot_pfn(pfn))) { - mark_mmio_spte(vcpu, sptep, gfn, pte_access); - return RET_PF_EMULATE; - } - if (is_shadow_present_pte(*sptep)) { /* * If we overwrite a PTE page pointer with a 2MB PMD, unlink @@ -2846,6 +2841,13 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, was_rmapped = 1; } + if (unlikely(is_noslot_pfn(pfn))) { + mark_mmio_spte(vcpu, sptep, gfn, pte_access); + if (flush) + kvm_flush_remote_tlbs_gfn(vcpu->kvm, gfn, level); + return RET_PF_EMULATE; + } + wrprot = make_spte(vcpu, sp, slot, pte_access, gfn, pfn, *sptep, prefetch, true, host_writable, &spte); -- 2.34.1

2 1

[PATCH OLK-6.6] KVM: x86/mmu: Drop/zap existing present SPTE even when creating an MMIO SPTE
by Zhang Kunbo 02 Apr '26

02 Apr '26

From: Sean Christopherson <seanjc(a)google.com> mainline inclusion from mainline-v7.0-rc6 commit aad885e774966e97b675dfe928da164214a71605 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/14045 CVE: CVE-2026-23401 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- When installing an emulated MMIO SPTE, do so *after* dropping/zapping the existing SPTE (if it's shadow-present). While commit a54aa15c6bda3 was right about it being impossible to convert a shadow-present SPTE to an MMIO SPTE due to a _guest_ write, it failed to account for writes to guest memory that are outside the scope of KVM. E.g. if host userspace modifies a shadowed gPTE to switch from a memslot to emulted MMIO and then the guest hits a relevant page fault, KVM will install the MMIO SPTE without first zapping the shadow-present SPTE. ------------[ cut here ]------------ is_shadow_present_pte(*sptep) WARNING: arch/x86/kvm/mmu/mmu.c:484 at mark_mmio_spte+0xb2/0xc0 [kvm], CPU#0: vmx_ept_stale_r/4292 Modules linked in: kvm_intel kvm irqbypass CPU: 0 UID: 1000 PID: 4292 Comm: vmx_ept_stale_r Not tainted 7.0.0-rc2-eafebd2d2ab0-sink-vm #319 PREEMPT Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 RIP: 0010:mark_mmio_spte+0xb2/0xc0 [kvm] Call Trace: <TASK> mmu_set_spte+0x237/0x440 [kvm] ept_page_fault+0x535/0x7f0 [kvm] kvm_mmu_do_page_fault+0xee/0x1f0 [kvm] kvm_mmu_page_fault+0x8d/0x620 [kvm] vmx_handle_exit+0x18c/0x5a0 [kvm_intel] kvm_arch_vcpu_ioctl_run+0xc55/0x1c20 [kvm] kvm_vcpu_ioctl+0x2d5/0x980 [kvm] __x64_sys_ioctl+0x8a/0xd0 do_syscall_64+0xb5/0x730 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x47fa3f </TASK> ---[ end trace 0000000000000000 ]--- Reported-by: Alexander Bulekov <bkov(a)amazon.com> Debugged-by: Alexander Bulekov <bkov(a)amazon.com> Suggested-by: Fred Griffoul <fgriffo(a)amazon.co.uk> Fixes: a54aa15c6bda ("KVM: x86/mmu: Handle MMIO SPTEs directly in mmu_set_spte()") Cc: stable(a)vger.kernel.org Signed-off-by: Sean Christopherson <seanjc(a)google.com> Conflicts: arch/x86/kvm/mmu/mmu.c [Context conflicts.] Signed-off-by: Zhang Kunbo <zhangkunbo(a)huawei.com> --- arch/x86/kvm/mmu/mmu.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index c54c8385b16d..2837f83b807d 100644 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -2914,12 +2914,6 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, bool prefetch = !fault || fault->prefetch; bool write_fault = fault && fault->write; - if (unlikely(is_noslot_pfn(pfn))) { - vcpu->stat.pf_mmio_spte_created++; - mark_mmio_spte(vcpu, sptep, gfn, pte_access); - return RET_PF_EMULATE; - } - if (is_shadow_present_pte(*sptep)) { /* * If we overwrite a PTE page pointer with a 2MB PMD, unlink @@ -2939,6 +2933,14 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, was_rmapped = 1; } + if (unlikely(is_noslot_pfn(pfn))) { + vcpu->stat.pf_mmio_spte_created++; + mark_mmio_spte(vcpu, sptep, gfn, pte_access); + if (flush) + kvm_flush_remote_tlbs_gfn(vcpu->kvm, gfn, level); + return RET_PF_EMULATE; + } + wrprot = make_spte(vcpu, sp, slot, pte_access, gfn, pfn, *sptep, prefetch, true, host_writable, &spte); -- 2.34.1

2 1

[PATCH OLK-6.6] KVM: x86/mmu: Drop/zap existing present SPTE even when creating an MMIO SPTE
by Zhang Kunbo 02 Apr '26

02 Apr '26

From: Sean Christopherson <seanjc(a)google.com> stable inclusion from stable-v7.0-rc6 commit aad885e774966e97b675dfe928da164214a71605 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/14045 CVE: CVE-2026-23401 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- When installing an emulated MMIO SPTE, do so *after* dropping/zapping the existing SPTE (if it's shadow-present). While commit a54aa15c6bda3 was right about it being impossible to convert a shadow-present SPTE to an MMIO SPTE due to a _guest_ write, it failed to account for writes to guest memory that are outside the scope of KVM. E.g. if host userspace modifies a shadowed gPTE to switch from a memslot to emulted MMIO and then the guest hits a relevant page fault, KVM will install the MMIO SPTE without first zapping the shadow-present SPTE. ------------[ cut here ]------------ is_shadow_present_pte(*sptep) WARNING: arch/x86/kvm/mmu/mmu.c:484 at mark_mmio_spte+0xb2/0xc0 [kvm], CPU#0: vmx_ept_stale_r/4292 Modules linked in: kvm_intel kvm irqbypass CPU: 0 UID: 1000 PID: 4292 Comm: vmx_ept_stale_r Not tainted 7.0.0-rc2-eafebd2d2ab0-sink-vm #319 PREEMPT Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 RIP: 0010:mark_mmio_spte+0xb2/0xc0 [kvm] Call Trace: <TASK> mmu_set_spte+0x237/0x440 [kvm] ept_page_fault+0x535/0x7f0 [kvm] kvm_mmu_do_page_fault+0xee/0x1f0 [kvm] kvm_mmu_page_fault+0x8d/0x620 [kvm] vmx_handle_exit+0x18c/0x5a0 [kvm_intel] kvm_arch_vcpu_ioctl_run+0xc55/0x1c20 [kvm] kvm_vcpu_ioctl+0x2d5/0x980 [kvm] __x64_sys_ioctl+0x8a/0xd0 do_syscall_64+0xb5/0x730 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x47fa3f </TASK> ---[ end trace 0000000000000000 ]--- Reported-by: Alexander Bulekov <bkov(a)amazon.com> Debugged-by: Alexander Bulekov <bkov(a)amazon.com> Suggested-by: Fred Griffoul <fgriffo(a)amazon.co.uk> Fixes: a54aa15c6bda ("KVM: x86/mmu: Handle MMIO SPTEs directly in mmu_set_spte()") Cc: stable(a)vger.kernel.org Signed-off-by: Sean Christopherson <seanjc(a)google.com> Conflicts: arch/x86/kvm/mmu/mmu.c [Context conflicts.] Signed-off-by: Zhang Kunbo <zhangkunbo(a)huawei.com> --- arch/x86/kvm/mmu/mmu.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index c54c8385b16d..2837f83b807d 100644 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -2914,12 +2914,6 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, bool prefetch = !fault || fault->prefetch; bool write_fault = fault && fault->write; - if (unlikely(is_noslot_pfn(pfn))) { - vcpu->stat.pf_mmio_spte_created++; - mark_mmio_spte(vcpu, sptep, gfn, pte_access); - return RET_PF_EMULATE; - } - if (is_shadow_present_pte(*sptep)) { /* * If we overwrite a PTE page pointer with a 2MB PMD, unlink @@ -2939,6 +2933,14 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot, was_rmapped = 1; } + if (unlikely(is_noslot_pfn(pfn))) { + vcpu->stat.pf_mmio_spte_created++; + mark_mmio_spte(vcpu, sptep, gfn, pte_access); + if (flush) + kvm_flush_remote_tlbs_gfn(vcpu->kvm, gfn, level); + return RET_PF_EMULATE; + } + wrprot = make_spte(vcpu, sp, slot, pte_access, gfn, pfn, *sptep, prefetch, true, host_writable, &spte); -- 2.34.1

2 1