From: Qing Luo refresh_signalling_expectation() converts the expiry from a SIP REGISTER 2xx response with "expires * HZ" in 32-bit arithmetic and stores it in a u32. sip_strtouint() only rejects values above UINT_MAX, so the answering registrar controls the input. Expectations are reaped with "(__s32)(exp->timeout - nfct_time_stamp) <= 0". A wrapped product can only cut the lifetime short, never extend it: at HZ=1000 an expiry of 2147484 seconds (24.9 days) crosses 2^31, so the expectation is born expired and registration refreshes break, while 4294968 seconds wraps to 704 jiffies and the expectation is reaped almost immediately. Add nf_ct_secs_to_jiffies(), which multiplies in u64 and saturates at INT_MAX so the signed expiry test stays positive, and use it for the SIP expectation refresh. Oversized expiries now saturate to the INT_MAX jiffies ceiling - about 24.8 days at HZ=1000 and 248 days at HZ=100 - which is the same maximum a registrar could already request with the largest non-wrapping value; the longest lifetime does not change. Sweeping all 2^32-1 inputs at HZ=1000, the helper never wraps, never yields a value that reads as expired, and matches the old expression up to INT_MAX / HZ. Fixes: 0f32a40fc91a ("[NETFILTER]: nf_conntrack_sip: create signalling expectations") Cc: stable@vger.kernel.org Signed-off-by: Qing Luo --- v2: - Clarified in the commit message that this patch introduces the nf_ct_secs_to_jiffies() helper, which the subsequent H.323 patch depends on. - Added a note about the stable backporting dependency for the helper. - Addressed the Sashiko AI review comments. v1: https://lore.kernel.org/all/20261104085925.773471-1-l1138897701@163.com/ --- include/net/netfilter/nf_conntrack.h | 10 ++++++++++ net/netfilter/nf_conntrack_sip.c | 3 ++- 2 files changed, 12 insertions(+), 1 deletion(-) diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h index c39425e54d87..b32716084ea8 100644 --- a/include/net/netfilter/nf_conntrack.h +++ b/include/net/netfilter/nf_conntrack.h @@ -210,6 +210,16 @@ bool nf_ct_get_tuplepr(const struct sk_buff *skb, unsigned int nhoff, void __nf_ct_refresh_acct(struct nf_conn *ct, enum ip_conntrack_info ctinfo, u32 extra_jiffies, unsigned int bytes); +/* Convert a timeout in seconds to jiffies, saturating at INT_MAX instead of + * wrapping the "secs * HZ" product in u32. + */ +static inline u32 nf_ct_secs_to_jiffies(u32 secs) +{ + u64 delta = (u64)secs * HZ; + + return delta > INT_MAX ? INT_MAX : (u32)delta; +} + /* Refresh conntrack for this many jiffies and do accounting */ static inline void nf_ct_refresh_acct(struct nf_conn *ct, enum ip_conntrack_info ctinfo, diff --git a/net/netfilter/nf_conntrack_sip.c b/net/netfilter/nf_conntrack_sip.c index 64bc440b1181..e2fd97f53386 100644 --- a/net/netfilter/nf_conntrack_sip.c +++ b/net/netfilter/nf_conntrack_sip.c @@ -891,7 +891,8 @@ static int refresh_signalling_expectation(struct nf_conn *ct, exp->tuple.dst.protonum != proto || exp->tuple.dst.u.udp.port != port) continue; - WRITE_ONCE(exp->timeout, nfct_time_stamp + (expires * HZ)); + WRITE_ONCE(exp->timeout, + nfct_time_stamp + nf_ct_secs_to_jiffies(expires)); WRITE_ONCE(exp->flags, exp->flags & ~NF_CT_EXPECT_INACTIVE); found = 1; break; -- 2.25.1