CVE-2026-18417
MediumCVSS 6.5Exploitation Probability (EPSS)
Low risk6th percentile - higher than 6% of all known CVEs
Summary
In Zephyr RTOS, a vulnerability exists in the BSD socket layer where an asynchronous error was stored in the user_data field of the network context, which for listening TCP sockets is used to store a pointer to the parent context. In v4.3.0, after an interface-down event, the accept callback was not disarmed, leading to reuse of the poisoned value as a pointer and writes to an invalid address, causing a kernel fatal error (device crash or reset).
Risk Assessment
An attacker who can repeatedly force a network interface down (e.g., by disrupting a wireless link) or with local/physical access can cause a denial of service (device crash or reset). The vulnerability does not allow direct code execution but can be used to destabilize the system.
Recommendation
Immediately update Zephyr RTOS to a version containing the fix (v4.3.1 or later, or a version with commit 269cb8823d3 / 913fae5169425550f2364655298fceb79b320066). If updating is not possible, restrict access to network interfaces and monitor link-down events.
Other vulnerabilities in Zephyr RTOS
See all- CVE-2026-19185High
A vulnerability was found in the system-call verifier for i3c_do_ccc() in Zephyr RTOS, which fails to validate per-target data pointers in the i3c_ccc_payload structure. The verifier also operates on the caller's live structure, allowing another thread to modify fields between check and use. This allows an unprivileged thread to write to arbitrary kernel addresses (via read CCC) or disclose kernel memory (via write CCC), leading to privilege escalation.
- CVE-2026-18747Medium
A vulnerability in the SMP-over-console transport in Zephyr RTOS allows an unauthenticated attacker to trigger a buffer length underflow by sending a single 7-byte frame to the management console. This leads to an out-of-bounds read, potentially causing a denial of service on the MCUmgr thread or memory disclosure.
- CVE-2026-16513High
A vulnerability in the RTIO syscall verifier in Zephyr RTOS (before v4.3.0) allows an unprivileged user thread to write a pointer to controlled data at an arbitrary kernel address. The missing K_SYSCALL_MEMORY_WRITE check for the handle output parameter enables privilege escalation to kernel mode.
- CVE-2026-16514Medium
The gptp_mi_qualify_announce() function in Zephyr's gPTP subsystem processes received Announce messages by comparing clock identities. The comparison loop uses an attacker-controlled step count (steps_removed) instead of the actual number of TLV entries, leading to an out-of-bounds read. A single frame can cause about 2KB of overread, potentially crashing the networking RX thread and causing denial of service.
- CVE-2026-16512Low
The gptp_handle_msg() function in Zephyr RTOS does not validate the received frame length before dereferencing the gPTP header pointer. A truncated frame can lead to out-of-bounds reads, disclosing stale memory contents.
- CVE-2026-14696Medium
A vulnerability in Zephyr RTOS causes network buffer (net_pkt) leaks when Ethernet bridging is enabled (CONFIG_NET_ETHERNET_BRIDGE). Frames with unrecognized EtherType that are also delivered to the local stack are marked as NET_OK, which is interpreted as consumed, but they are never freed. An attacker on the L2 segment can flood broadcast/multicast frames with arbitrary EtherType, exhausting the RX buffer pool and causing a persistent denial of service until reboot.
- CVE-2026-13480Low
A vulnerability in the LoRaWAN TS004 Fragmented Data Block Transport handler frag_transport_package_callback() in Zephyr RTOS (subsys/lorawan/services/frag_transport.c) lacks validation of remaining payload bytes before each access. An attacker with session keys can cause an out-of-bounds read (up to ~232 bytes) past the RxPayload buffer, copying adjacent static memory into decoder buffers and the FUOTA flash image.
- CVE-2026-13213Medium
A vulnerability in the Hearing Access Service (HAS) in Zephyr RTOS allows a remote (Bluetooth) crash of the HAS peripheral. The issue occurs when a previously bonded client reconnects during the startup window before the application registers the service, leading to an assertion or NULL pointer dereference.
- CVE-2026-12633High
Vulnerability in IPv6 neighbor-discovery code (subsys/net/ip/ipv6_nbr.c) in Zephyr RTOS. The handle_ra_6co() function does not bound the context_len field from the 6CO option in Router Advertisements, leading to a buffer overflow and out-of-bounds memset. An attacker on the same link can send a crafted packet causing a system crash.
- CVE-2026-9771High
A vulnerability in the flash_copy() system call in Zephyr RTOS (with CONFIG_USERSPACE enabled) allows an unprivileged thread to pass forged pointers to flash device structures. Lack of device object validation (src_dev, dst_dev) before dereferencing enables arbitrary code execution in supervisor mode, leading to privilege escalation outside the userspace sandbox.
Original NVD description (English source)
The native BSD-socket layer recorded a pending asynchronous socket error by type-punning it into struct net_context's void user_data field (ctx->user_data = INT_TO_POINTER(-status) in zsock_accepted_cb(), zsock_received_cb(), zsock_connected_cb() and zsock_close_ctx() in subsys/net/lib/sockets/sockets_inet.c), reading it back with POINTER_TO_INT(). That same field is owned by the network stack for listening TCP contexts: net_tcp_accept() stores the parent context pointer there and the TCP core passes it back to the registered accept callback. A failed accept therefore left a small integer (an errno value) where the stack expected a struct net_context . When the network interface carrying a listening TCP socket goes down, close_tcp_conn() in subsys/net/ip/tcp.c invokes the accept callback with -ENETDOWN and the context's user_data. In v4.3.0 the callback was not disarmed afterwards, so a second interface-down event forwarded the previously stored errno to zsock_accepted_cb(), which dereferenced it as the parent context and performed several stores through it (sock_set_error()'s read-modify-write of socket_data, k_fifo_cancel_wait(&parent->recv_q)) — the crash described in the fix's commit message. v4.3.1 and v4.4.x carry a later change clearing conn->accept_cb after the error callback (269cb8823d3 on the v4.3 branch, 913fae5169425550f2364655298fceb79b320066 on main), which closes that repeat path; on those releases the poisoned cookie remains reachable only by a narrower race, a handshake completing alongside the interface-down still passing the stale cookie to k_fifo_put(&parent->accept_q, ...), and by getsockopt(SO_ERROR), which reads the field back unconditionally. On v4.3.0 an application that keeps a listening TCP socket open across repeated link-down events is sufficient to reach the defect; the triggering condition is a network-interface state change, not attacker-supplied packet data, so the practical attacker is one able to force the link down repeatedly (for example an adjacent attacker disrupting a wireless link) or one with local/physical access. Because both the faulting address and the stored data are fixed small constants derived from the errno value, the outcome is a wild-pointer access leading to a kernel fatal error — a denial of service (device crash or reset) rather than an attacker-directed memory corruption. The fix stores the pending error in a dedicated net_context.sock_error field and converts every producer and consumer to sock_set_error()/sock_get_error(), leaving user_data untouched. As a side effect it also stops getsockopt(SO_ERROR) — which is evaluated unconditionally — from returning the kernel address held in user_data to a userspace application.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS