CVE-2026-99676: net: restore refcount ordering for lockless socket access
In the Linux kernel, the following vulnerability has been resolved:
net: restore refcount ordering for lockless socket access
Before commit fb041bb7c0a9 ("locking/refcount: Consolidate implementations of refcountt"), default builds implemented refcountincnotzero() using atomicaddunless(), which provided full memory ordering on success. The commit removed that ordering on weakly ordered architectures, allowing accesses after a successful reference acquisition to be reordered before it.
Use refcountincnotzeroacquire() in the IPv4 and IPv6 established lookups and in inettwskpurge(), where the socket is validated again after taking a reference.
bpfskassign() can receive a socket pointer obtained through BTF without a reference. Use acquire ordering when it takes its own reference for a refcounted socket.
Use refcountsetrelease() when setting refcounts in the SYN cookie, TCP Fast Open, request migration, and TIMEWAIT paths. Keep the existing TIMEWAIT smpwmb(), which separately orders the refcount store before ehash publication.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 5.5 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.12.118 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.18.64 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 7.3.5 - Upgrade
Upgrade
debian/linuxto a version that resolves this vulnerability.Fixed in 7.3.5-1
Event History
Frequently Asked Questions
Which systems are exposed to this issue?
Systems running Linux kernels on weakly ordered architectures are affected by the loss of memory ordering described in the fix. The affected paths involve IPv4 and IPv6 established socket lookups, TIME_WAIT socket purging, BPF socket assignment, SYN cookies, TCP Fast Open, and request migration.
Does exploitation require an attacker to have local BPF access?
No. The fix covers network socket lookup and lifecycle paths in addition to bpf_sk_assign(). The provided information does not specify a particular attacker access level or trigger sequence.
How can I determine whether my kernel includes the fix?
Check whether your kernel contains the stable commits referenced for this issue. The remediation replaces relevant reference acquisitions with acquire-ordered operations and uses release-ordered refcount initialization in the affected socket lifecycle paths.