CVE-2026-99751: bpf: Fix skb_pull() BUG in __bpf_redirect_neigh()

Published Nov 20, 2026
·
Updated

In the Linux kernel, the following vulnerability has been resolved:

bpf: Fix skbpull() BUG in bpfredirectneigh()

The issue: This is a follow-up to the short link-layer header fixes from Eric. After vlan popeth, a privileged SCHEDACT BPF program can call bpfskbchangehead(skb, 1, 0) and then bpfredirectneigh().

bpfredirectneigh() calls skbpull(skb, sizeof(ethh)) after checking only that macheader < networkheader. Nothing verifies that the MAC header is long enough, that it is linear, or that the IP header the redirect goes on to read is linear either.

A SCHEDACT BPF program can do:

bpfskbchangehead(skb, 1, 0); bpfredirectneigh(lo, NULL, 0, 0);

after vlan popeth. The skb then has maclen == 1 and only one byte of linear data; skbpull(ETHHLEN) trips the skb->len < skb->datalen BUG in skbpull():

kernel BUG at ./include/linux/skbuff.h:2839! RIP: skbpull+0x40/0x50 Call Trace: skbdoredirect -> packetxmit -> packetsendmsg Kernel panic - not syncing: Fatal exception in interrupt

The v4 and v6 paths read iphdr() and ipv6hdr() as soon as the MAC header is gone. On the AFPACKET frame below that leaves headlen == 0 after the pull, so the network header sits at skb->tail and the route lookup reads ip4h->protocol, ip4h->saddr and ip4h->daddr past the linear area.

The Fix: Use skbnetworkoffset(), which is the MAC header span, instead of skb->maclen. maclen is zero for locally generated tc egress and netkit packets, while the span still rejects the one-byte header produced by popeth + bpfskbchangehead(skb, 1, 0). The single check also covers the macheader >= networkheader test it replaces.

Use pskbmaypull() to make the Ethernet header linear before it is read or pulled, and to make the IP header linear in both the v4 and the v6 path before iphdr()/ipv6hdr() is read. The header pointers are taken after the pull, since pskbmaypull() may reallocate skb->head.

Conditions to recreate the bug: On x86-64 with 4 KiB pages, enable CONFIGPACKET, CONFIGBPFSYSCALL, CONFIGNETSCHINGRESS, CONFIGNETCLSMATCHALL, CONFIGNETACTVLAN and CONFIGNETACTBPF. Run as root with BPF loading, network administration and raw-packet privileges. Bring lo up and attach a SCHEDACT program that calls bpfskbchangehead(skb, 1, 0), then returns bpfredirectneigh(skb->ifindex, NULL, 0, 0):

ip link set lo up tc qdisc add dev lo clsact tc filter add dev lo egress matchall action vlan popeth pipe \ action bpf object-file neighchghead.o section action

Send an untagged ETHPIP frame of ETHHLEN + 8192 bytes through an AFPACKET/SOCKRAW socket bound to lo, without PACKETVNETHDR or PACKETQDISCBYPASS. Use a unicast destination MAC. The sender's skb has a 14-byte linear Ethernet prefix and a fragmented payload; popeth removes that prefix and changehead restores only one linear byte. The unfixed neighbor redirect then tries to pull 14 bytes and BUGs.

Sending the same frame without popeth and changehead leaves the IP header in the frags, which is the route lookup reading past skb->tail.

Affected Software

7 affected componentsFixes available
linux/kernel>=5.10.0<6.12.118
6.12.118
linux/kernel>=6.13.0<6.18.64
6.18.64
linux/kernel>=6.19.0<7.3.5
7.3.5
debian/linux*
debian/linux*
debian/linux*
debian/linux<=6.1.176-1, <=6.1.192-2, <=6.12.107-1, <=6.12.117-1, <=7.3.4-1
7.3.5-1

Remediation

Recommended actions to resolve this vulnerability, in priority order.

  1. Upgrade

    Upgrade linux/Kernel to a version that resolves this vulnerability.

    Fixed in 6.12.118
  2. Upgrade

    Upgrade linux/Kernel to a version that resolves this vulnerability.

    Fixed in 6.18.64
  3. Upgrade

    Upgrade linux/Kernel to a version that resolves this vulnerability.

    Fixed in 7.3.5
  4. Upgrade

    Upgrade debian/linux to a version that resolves this vulnerability.

    Fixed in 7.3.5-1

Event History

Nov 20, 2026
CVE Published
via MITRE·05:51 AM
Data Sourced
via MITRE·05:51 AM
Description
Data Sourced
via OSV·05:51 AM
DescriptionAffected Software
Data Sourced
via NVD·06:18 AM
Description

Frequently Asked Questions

1

What level of access is needed to trigger this issue?

An attacker needs the ability to run a privileged SCHED_ACT BPF program. The described trigger requires that program to operate after vlan pop_eth and invoke BPF helpers that alter the skb head and redirect to a neighbor.

2

What is the practical impact of successful exploitation?

The malformed skb state can cause skb_pull() to hit a kernel BUG. The reported outcome is a kernel panic, causing a denial of service.

3

How can I assess whether deployed BPF programs may trigger it?

Review privileged SCHED_ACT BPF programs that run after vlan pop_eth. The relevant pattern is bpf_skb_change_head(skb, 1, 0) followed by bpf_redirect_neigh(), which can leave only one byte of linear MAC-header data before the redirect path pulls an Ethernet header.

4

Are IP redirect paths involved?

Yes. After the MAC header is removed, both IPv4 and IPv6 redirect paths read their respective IP headers. The issue also involves cases where those headers are not linear, including the described AF_PACKET frame condition.

Contact

SecAlerts Pty Ltd.
132 Wickham Terrace
Fortitude Valley,
QLD 4006, Australia
info@secalerts.co
By using SecAlerts services, you agree to our services end-user license agreement. This website is safeguarded by reCAPTCHA and governed by the Google Privacy Policy and Terms of Service. All names, logos, and brands of products are owned by their respective owners, and any usage of these names, logos, and brands for identification purposes only does not imply endorsement. If you possess any content that requires removal, please get in touch with us.
© 2026 SecAlerts Pty Ltd.
ABN: 70 645 966 203, ACN: 645 966 203