CVE-2025-40347

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: enetc: fix the deadlock of enetc_mdio_lock<br /> <br /> After applying the workaround for err050089, the LS1028A platform<br /> experiences RCU stalls on RT kernel. This issue is caused by the<br /> recursive acquisition of the read lock enetc_mdio_lock. Here list some<br /> of the call stacks identified under the enetc_poll path that may lead to<br /> a deadlock:<br /> <br /> enetc_poll<br /> -&gt; enetc_lock_mdio<br /> -&gt; enetc_clean_rx_ring OR napi_complete_done<br /> -&gt; napi_gro_receive<br /> -&gt; enetc_start_xmit<br /> -&gt; enetc_lock_mdio<br /> -&gt; enetc_map_tx_buffs<br /> -&gt; enetc_unlock_mdio<br /> -&gt; enetc_unlock_mdio<br /> <br /> After enetc_poll acquires the read lock, a higher-priority writer attempts<br /> to acquire the lock, causing preemption. The writer detects that a<br /> read lock is already held and is scheduled out. However, readers under<br /> enetc_poll cannot acquire the read lock again because a writer is already<br /> waiting, leading to a thread hang.<br /> <br /> Currently, the deadlock is avoided by adjusting enetc_lock_mdio to prevent<br /> recursive lock acquisition.