CVE-2026-74691
HighCVSS 8.8Exploitation Probability (EPSS)
Low risk16th percentile - higher than 16% of all known CVEs
Summary
In the Linux kernel's Thunderbolt network driver, a vulnerability was found in the teardown sequence: DMA rings are stopped and frame buffers freed before DMA paths are deactivated. This causes in-flight data to have nowhere to drain, so the 'pending' bit never clears, leading to timeouts and link failure.
Risk Assessment
The risk is that repeated teardowns of the Thunderbolt interface can completely hang the XDomain controller, requiring a system reboot. An attacker with network access could intentionally trigger such events, causing denial of service (DoS).
Recommendation
Apply the kernel patch that reverses the teardown order (deactivate DMA paths first, then stop rings). Update the system to a kernel version that includes this fix.
Other vulnerabilities in Linux kernel
See all- CVE-2026-99215Unknown
In the Linux kernel power supply subsystem, a use-after-free vulnerability exists. If a power supply is unregistered while its external_power_changed() callback is still running, the driver may free data that the callback is still accessing, leading to a kernel crash (NULL pointer dereference).
- CVE-2026-99210Unknown
In the Linux kernel accel/rocket driver, rocket_probe() stores the return value of rocket_device_init() in the file-scoped rdev pointer before checking whether it is an error pointer. On failure rdev remains an ERR_PTR, and a later probe can skip initialization and dereference the error pointer when reading rdev->num_cores.
- CVE-2026-99209Unknown
In the Linux kernel drm/xe driver, xe_evict_flags() reads tbo->resource->mem_type, but a BO can reach eviction with no ttm.resource, e.g. an imported dma-buf whose resource was freed during teardown. This causes a NULL pointer dereference and a kernel fault.
- CVE-2026-99208Unknown
In the Linux kernel drm/xe/oa driver, ttm_bo_type_kernel BOs are not guaranteed to be zeroed on allocation and may contain data from previously freed pages. Since the OA buffer is mmap'able by userspace, this can expose stale kernel memory contents.
- CVE-2026-99207Unknown
In the Linux kernel's scsi: target subsystem, target_do_delayed_work() acquired delayed_cmd_lock with a plain spin_lock() without disabling interrupts, while all other acquisition sites use spin_lock_irqsave(). Since commands can now complete in interrupt context (complete_type=1 with direct_compl_supp), this can cause a deadlock when a completion arrives on the same CPU where the work item holds the lock.
- CVE-2026-99205Unknown
In the Linux kernel, a race exists in sched_change_end() with __sched_setscheduler() for queued but non-running tasks. When a task's scheduling class changes (e.g., from SCHED_OTHER to SCHED_RR), rq->next_class is not updated and the need-resched indicator is not set, causing the CPU to idle despite a runnable task being present.
- CVE-2026-99200Unknown
In the Linux kernel, the hisi_femac network driver (net: hisilicon) assigns netdev_ops only after connecting the PHY. With CONFIG_NET_SHAPER enabled, phy_attach_direct() reads uninitialized netdev_ops, causing a kernel panic from a NULL pointer dereference.
- CVE-2026-99199Unknown
In the Linux kernel, the hip04_eth network driver (net: hisilicon) calls of_phy_connect() before ndev->netdev_ops is assigned. With CONFIG_NET_SHAPER enabled, phy_attach_direct() reads uninitialized netdev_ops, causing a kernel panic from a NULL pointer dereference.
- CVE-2026-99198Unknown
In the Linux kernel, the owl_emac network driver (net: ethernet: actions) calls owl_emac_phy_init(), which connects the PHY, before netdev->netdev_ops is assigned. With CONFIG_NET_SHAPER enabled, phy_attach_direct() reads uninitialized netdev_ops, causing a kernel panic from a NULL pointer dereference.
- CVE-2026-99197Unknown
In the Linux kernel, the mv643xx_eth network driver connects the PHY before dev->netdev_ops is assigned. With CONFIG_NET_SHAPER enabled, phy_attach_direct() reads uninitialized netdev_ops, causing a kernel panic from a NULL pointer dereference.
Original NVD description (English source)
In the Linux kernel, the following vulnerability has been resolved: net: thunderbolt: Tear down DMA paths before stopping the rings tbnet_tear_down() stops both rings and frees their frame buffers before calling tb_xdomain_disable_paths(). tb_ring_stop() zeroes the ring's descriptor base and tbnet_free_buffers() unmaps and frees the pages the frames sit in, so by the time __tb_path_deactivate_hop() polls the hop's 'pending' bit, anything still in flight has nowhere to drain to. The teardown sequence has been in this order since the driver was added. The setup path has not: commit ff7cd07f3064 ("net: thunderbolt: Enable DMA paths only after rings are enabled") moved the path enable to the end of tbnet_connected_work() and documented why: /* Both logins successful so enable the rings, high-speed DMA * paths and start the network device queue. * * Note we enable the DMA paths last to make sure we have primed * the Rx ring before any incoming packets are allowed to * arrive. */ Teardown was never updated to match, so the rings and the paths now come down in the same order they go up instead of in reverse. On an ASMedia ASM4242 host router the 'pending' bit then never clears: every teardown burns the full 500 ms timeout and __tb_path_deactivate_hop() returns -ETIMEDOUT. Raising the timeout to 5 s does not help, so the hop is not slow to drain, it never drains at all. The failure is invisible above the thunderbolt core. __tb_path_deactivate_hops() is void and only calls tb_port_warn(); tb_path_deactivate(), tb_tunnel_deactivate() and __tb_disconnect_xdomain_paths() are void as well, and tb_disconnect_xdomain_paths() ends in an unconditional "return 0". So tb_xdomain_disable_paths() reports success and the netdev_warn() below it never fires. Repeated teardowns eventually take the XDomain control channel down, after which the peer node is gone and only a power cycle brings the controller back. Deactivating the paths first fixes it. Measured with kretprobes on a stock v6.17 tree with no other patches applied, on a link that was up and had just carried traffic: before: __tb_path_deactivate_hop() returns 0 for the first hop, then -ETIMEDOUT for the second 500335 us later after: 0 for both, 525 us apart Alternating the two orderings ABBA over three load levels, four teardowns per arm: every teardown failed before the change (21 of 21 that ran), none failed after (0 of 24). The before arms ran short because the link died partway through. The same split shows up when the interface is enslaved to a bond instead of just brought down, which is how I ran into this in the first place. Throughput and latency after the change are unchanged. Hosts whose routers drain the hop despite the stale descriptor base see no functional difference, since the paths end up deactivated either way.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS