CVE-2026-98369
UnknownSummary
A bug was fixed in the Linux kernel xfrm subsystem where xfrm_trans_reinject(), after being converted from a tasklet to a workqueue, ran in process context without an RCU read lock. This caused suspicious RCU usage warnings and potential validity issues for dst and netdevice references of packets queued through the workqueue.
Risk Assessment
It can lead to kernel instability, lockdep warnings, and in extreme cases use-after-free when processing xfrm packets. It affects systems using transport-mode IPsec.
Recommendation
Update the Linux kernel to a version containing this fix. If immediate update is not possible, consider limiting the use of transport-mode xfrm until the patch is deployed.
Other vulnerabilities in Linux kernel
See all- CVE-2026-98372Unknown
A vulnerability in the Linux kernel's xfrm IP-TFS module allows an out-of-bounds stack read in iptfs_skb_reset_frag_walk() due to missing fragment index validation. A crafted AGGFRAG packet can trigger this, causing a system crash or potential information leak.
- CVE-2026-98371Unknown
In the Linux kernel, a flaw in xfrm/iptfs handling of inner packet fragmentation occurs when the start of an inner packet is split across two outer packets such that fewer than 4 bytes land at the end of the first one. This bypasses IP header length validation, leading to a buffer overflow and kernel panic (DoS).
- CVE-2026-98370Unknown
In the Linux kernel, a use-after-free vulnerability was fixed in xfrm_state_netlink() when handling ALLOCSPI requests. xfrm_alloc_userspi() calls alloc_compat() again, but passes the original request skb and its header, which for compat requests causes the 228-byte compat xfrm_userspi_info to be interpreted as the 232-byte native layout and reads four bytes past the declared payload. A multicast clone of the request shares skb_shared_info and can observe that child, and xfrm_user_rcv_msg() frees it after the request handler returns, leading to a race and use-after-free.
- CVE-2026-98368Unknown
A vulnerability in the Linux kernel's ESP (IPsec) implementation was found on the out-of-place output path, where the skb frag array was mutated without downgrading managed zerocopy frags (SKBFL_MANAGED_FRAG_REFS). This breaks the managed-frag invariant, potentially leading to use-after-free of zerocopy pages and memory leaks.
- CVE-2026-98367Unknown
In the Linux kernel, a vulnerability was fixed in the RDMA/siw driver in siw_accept(). If siw_qp_modify() fails, the QP's state_lock is released before error path cleanup, allowing a concurrent ibv_modify_qp() to transition the QP to ERROR and free the cep structure, leading to use-after-free. The fix clears qp->cep and drops the association reference under the held write lock.
- CVE-2026-98366Unknown
In the Linux kernel's RDMA/rxe subsystem, a vulnerability exists where rxe_rereg_user_mr() reassigns the PD pointer before validating access flags. If validation fails, the MR remains in an inconsistent state, leading to a use-after-free of the PD memory.
- CVE-2026-98365Unknown
In the Linux kernel RDMA/rxe subsystem, mr_check_range() has an integer overflow vulnerability that allows a remote unauthenticated client to bypass memory range validation and trigger out-of-bounds access, causing a kernel oops. The fix rewrites the check in an overflow-safe form, returning -EINVAL for crafted requests.
- CVE-2026-98364Unknown
In the Linux kernel xfrm subsystem, during bundle creation, a missing reference to net_device leads to a use-after-free when a concurrent interface deletion occurs. The fix reads dst->dev via dst_dev_rcu() and keeps the RCU read-side critical section active until xfrm_fill_dst() takes the required references.
- CVE-2026-98363Unknown
In the Linux kernel firmware arm_scpi subsystem, scpi_dvfs_get_info() does not check the upper bound of the OPP count read from SCP firmware, which can lead to out-of-bounds read and incorrect sizing of the allocated OPP table. The fix rejects both zero and too-large OPP counts, returning -EINVAL.
- CVE-2026-98362Unknown
In the Linux kernel clk scpi subsystem, scpi_dvfs_recalc_rate() does not check the upper bound of the DVFS index returned by dvfs_get_idx(), which can lead to out-of-bounds read and treating garbage as clock frequency. The fix treats indexes greater than or equal to the OPP count as invalid and returns 0.
Original NVD description (English source)
In the Linux kernel, the following vulnerability has been resolved: xfrm: add missing rcu_read_lock(), skb_dst_force() and dev_hold() for xfrm_trans_reinject() syzbot reported a suspicious RCU usage warning in ip6_pkt_drop(): WARNING: suspicious RCU usage in ip6_pkt_drop include/net/addrconf.h:389 suspicious rcu_dereference_check() usage! Call Trace: __in6_dev_get_safely include/net/addrconf.h:389 [inline] ip6_pkt_drop+0x596/0x610 net/ipv6/route.c:4620 ip6_pkt_discard+0x1c/0x30 net/ipv6/route.c:4651 xfrm_trans_reinject+0x324/0x630 net/xfrm/xfrm_input.c:806 process_one_work kernel/workqueue.c:3322 [inline] process_scheduled_works+0xa8e/0x14e0 kernel/workqueue.c:3405 worker_thread+0xa47/0xfb0 kernel/workqueue.c:3486 When commit 4f4920669d21 ("xfrm: Reinject transport-mode packets through workqueue") converted xfrm_trans_reinject from a tasklet to a workqueue, the reinjection loop ceased running in softirq context. Workqueue workers run in process context where local_bh_disable() does not enter an RCU read-side critical section under CONFIG_PREEMPT_RCU. Because finish callbacks (such as ip6_rcv_finish) expect to run under an RCU read lock (performing route lookups, l3mdev lookups, and accessing RCU-protected data structures), invoking them in workqueue context without rcu_read_lock() triggers RCU lockdep warnings. Furthermore, packets queued to the workqueue via xfrm_trans_queue_net() may carry non-refcounted (noref) dst entries (e.g. from ip_route_input_noref). Additionally, on netdevice unregistration, dst_dev_put() replaces dst->dev with blackhole_netdev, so dst entries do not keep skb->dev alive while queued in the workqueue. Fix these issues by: 1. Calling skb_dst_force(skb) in xfrm_trans_queue_net() while still in the caller's RCU section to ensure dst is reference-counted before queuing. 2. Holding a reference on skb->dev via dev_hold()/dev_put() across workqueue deferral so skb->dev remains valid during finish() callback processing. 3. Acquiring rcu_read_lock() around the finish callback invocation loop in xfrm_trans_reinject().
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

