{"id":"CVE-2026-98809","title":"Bluetooth: hci_sync: don't drain cmd_sync backlog on unregister","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nBluetooth: hci_sync: don't drain cmd_sync backlog on unregister\n\nhci_unregister_dev() disables cmd_work and cmd_timer, then calls\nhci_cmd_sync_clear(), whose cancel_wor…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < 143bde97c4101ae68eb656c87e620c45b07a3090","Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < 7f16443a30f4e86d84ae4faf057d333b41ddb757","Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < e8ba8c9c74d1035042cdf6cb5adf8f59b9b1271c","Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < b1f24c1ab52345e810c42a1a81286cb5a7c92252","Linux 5.17"],"published":"2026-10-18","updated":"2026-10-19","sourceUpdated":"2026-10-19T09:26:02.636Z","source":"CVEORG","sourceUrl":"https://www.cve.org/CVERecord?id=CVE-2026-98809","references":[{"url":"https://git.kernel.org/stable/c/143bde97c4101ae68eb656c87e620c45b07a3090"},{"url":"https://git.kernel.org/stable/c/7f16443a30f4e86d84ae4faf057d333b41ddb757"},{"url":"https://git.kernel.org/stable/c/e8ba8c9c74d1035042cdf6cb5adf8f59b9b1271c"},{"url":"https://git.kernel.org/stable/c/b1f24c1ab52345e810c42a1a81286cb5a7c92252"}],"tags":["cve.org"],"ingestedAt":"2026-10-18T16:47:32.478Z","slug":"CVE-2026-98809","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nBluetooth: hci_sync: don't drain cmd_sync backlog on unregister\n\nhci_unregister_dev() disables cmd_work and cmd_timer, then calls\nhci_cmd_sync_clear(), whose cancel_work_sync() waits for\nhci_cmd_sync_work() to return. That worker dequeues and runs every\nentry on cmd_sync_work_list. With cmd_work disabled nothing reaches\nthe controller, so each queued HCI command waits the full\nHCI_CMD_TIMEOUT (2s).\n\nThe backlog has no bound. Userspace can keep queuing MGMT commands\nsuch as MGMT_OP_GET_CLOCK_INFO against a controller that doesn't\nanswer. Closing /dev/vhci then blocks in vhci_release() for\nbacklog * 2s:\n\n  INFO: task syz-executor:5749 blocked for more than 143 seconds.\n   cancel_work_sync\n   hci_cmd_sync_clear\n   hci_unregister_dev\n   vhci_release\n\nIn a local reproduction the backlog held more than 5000 entries,\nwhich comes to hours of hang.\n\nStop the worker from taking new entries once HCI_UNREGISTER is set,\nand wake any request still waiting with -ENODEV before cancelling\nthe work. The entries left over are destroyed with -ECANCELED by the\nexisting sweep in hci_cmd_sync_clear(). They stay on the list until\nthe work has stopped, so hci_cmd_sync_dequeue() and friends still see\nthem. A callback already running may still issue another command,\nwhich delays unregister by at most one timeout per remaining command,\nnot by the whole backlog.\n\n## Affected\n\n- `Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < 143bde97c4101ae68eb656c87e620c45b07a3090`\n- `Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < 7f16443a30f4e86d84ae4faf057d333b41ddb757`\n- `Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < e8ba8c9c74d1035042cdf6cb5adf8f59b9b1271c`\n- `Linux >= 008ee9eb8a11bcabf12c91771dd4f470b082bd44 < b1f24c1ab52345e810c42a1a81286cb5a7c92252`\n- `Linux 5.17`\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":[]}