git: 71f4eb518cd9 - stable/15 - pf: don't reject route-to'd too-large packets
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Sat, 10 Jan 2026 10:22:50 UTC
The branch stable/15 has been updated by kp:
URL: https://cgit.FreeBSD.org/src/commit/?id=71f4eb518cd9241b429afebe5333aa1d6b7b46c6
commit 71f4eb518cd9241b429afebe5333aa1d6b7b46c6
Author: Kristof Provost <kp@FreeBSD.org>
AuthorDate: 2025-12-26 09:58:59 +0000
Commit: Kristof Provost <kp@FreeBSD.org>
CommitDate: 2026-01-10 09:50:37 +0000
pf: don't reject route-to'd too-large packets
If we're sending a packet via pf_route()/pf_route6() we check for packet
size and potentially generate ICMP(6) packet too big messages. If we do,
don't consider this a rejected packet. That is, return PF_PASS and set
the mbuf to NULL rather than returning PF_DROP.
This matters for locally generated packets, because with PF_DROP we
can end up returning EACCES to userspace, causing the connection to
terminate. Instead, with PF_PASS and a NULL mbuf this is translated to
PFIL_CONSUMED, which does not return an error to userspace.
MFC after: 2 weeks
Sponsored by: Rubicon Communications, LLC ("Netgate")
(cherry picked from commit 2e7699355f08258365fb5f65d11ac297e20f78de)
---
sys/netpfil/pf/pf.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/sys/netpfil/pf/pf.c b/sys/netpfil/pf/pf.c
index 567615acdf0c..2cd3ee535b85 100644
--- a/sys/netpfil/pf/pf.c
+++ b/sys/netpfil/pf/pf.c
@@ -9343,7 +9343,8 @@ pf_route(struct pf_krule *r, struct ifnet *oifp,
ifp->if_mtu, pd->af, r, pd->act.rtableid);
}
SDT_PROBE1(pf, ip, route_to, drop, __LINE__);
- action = PF_DROP;
+ /* Return pass, so we return PFIL_CONSUMED to the stack. */
+ action = PF_PASS;
goto bad;
}
@@ -9665,7 +9666,8 @@ pf_route6(struct pf_krule *r, struct ifnet *oifp,
pf_send_icmp(m0, ICMP6_PACKET_TOO_BIG, 0,
ifp->if_mtu, pd->af, r, pd->act.rtableid);
}
- action = PF_DROP;
+ /* Return pass, so we return PFIL_CONSUMED to the stack. */
+ action = PF_PASS;
SDT_PROBE1(pf, ip6, route_to, drop, __LINE__);
goto bad;
}