{"id":"CVE-2026-98617","title":"eth: fbnic: use the Rx queue napi pointer to find the napi vector","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\neth: fbnic: use the Rx queue napi pointer to find the napi vector\n\nThe queue management ndos pick the napi vector for an Rx queue with:\n\tnv = fbn->napi[idx % fbn->num_n…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= da43127a8edc7498cce7b2554533f4027c618933 < 1fb569cf3bf97e552f465b2c719e36b48ae7c18b","Linux >= da43127a8edc7498cce7b2554533f4027c618933 < b5d9e9d4d0c13bc8b60d8d97e7a07fb25fea639e","Linux 6.18"],"published":"2026-10-18","updated":"2026-10-19","sourceUpdated":"2026-10-19T09:22:45.221Z","source":"CVEORG","sourceUrl":"https://www.cve.org/CVERecord?id=CVE-2026-98617","references":[{"url":"https://git.kernel.org/stable/c/1fb569cf3bf97e552f465b2c719e36b48ae7c18b"},{"url":"https://git.kernel.org/stable/c/b5d9e9d4d0c13bc8b60d8d97e7a07fb25fea639e"}],"tags":["cve.org"],"ingestedAt":"2026-10-18T15:45:12.240Z","slug":"CVE-2026-98617","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\neth: fbnic: use the Rx queue napi pointer to find the napi vector\n\nThe queue management ndos pick the napi vector for an Rx queue with:\n\tnv = fbn->napi[idx % fbn->num_napi];\n\nThe issue is this is only correct in the cases where there are no\nstandalone Tx vectors. In those cases we were allocating the Tx vectors\nfirst and then the Rx so the queues would be pointing to Tx NAPI vectors\ninstead of the Rx ones.\n\nThe mapping the ndos want is already recorded.  fbnic_set_netif_napi()\npublishes it with netif_queue_set_napi(), which stores the napi pointer\nin netdev_rx_queue.napi, and fbnic_reset_netif_napi() clears it again.\nBoth run under the netdev instance lock that the queue management ndos\nalso hold, so the pointer can be read directly.\n\nUse it and drop the divide.  The pointer is NULL exactly while the\ndatapath is down, so fbnic_queue_mem_alloc() can reject that case rather\nthan reaching into freed state: netdev_rx_queue_restart() calls it\nbefore it tests netif_running(), and fbnic_pm_suspend() leaves\nnetif_running() true across a PCIe recovery that never completes, so a\nqueue restart can arrive after fbnic_stop() has freed the rings and the\nvectors.  fbnic_stop() clears the association in\nfbnic_reset_netif_queues() before fbnic_free_napi_vectors(), so the\nNULL is always published first.  fbnic_queue_start() and\nfbnic_queue_stop() need no check of their own, as\nnetdev_rx_queue_reconfig() only reaches them once fbnic_queue_mem_alloc()\nhas succeeded under the same instance lock.\n\n## Affected\n\n- `Linux >= da43127a8edc7498cce7b2554533f4027c618933 < 1fb569cf3bf97e552f465b2c719e36b48ae7c18b`\n- `Linux >= da43127a8edc7498cce7b2554533f4027c618933 < b5d9e9d4d0c13bc8b60d8d97e7a07fb25fea639e`\n- `Linux 6.18`\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":3,"depthScoreParts":{"impact":2.8,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}