{"id":"CVE-2026-98766","published":"2026-10-18T16:17:30.440","lastModified":"2026-10-19T10:19:40.727","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nusb: gadget: aspeed-vhub: cancel wake work on device removal\n\nwake_work is armed from the gadget .wakeup callback to resume suspended\ndownstream ports, and it is never cancelled in ast_vhub_remove(), so a\nwork item queued before or during removal can run after devm has freed\nvhub and its ports, a use-after-free.\n\nCancelling the work alone is not sufficient: usb_gadget_wakeup() takes\nno lock and the unbind path never clears wakeup_en, so a remote-wakeup\nrequest in flight on another CPU can re-arm the work after the cancel\nand before vhub is freed.\n\nast_vhub_del_dev() already clears d->registered under vhub->lock before\nunregistering the gadget. Refuse the wakeup in ast_vhub_udc_wakeup()\nonce d->registered is clear: the check and the schedule_work() are then\natomic against the unbind, so a wakeup that passed before the flag was\ncleared is drained by the cancel_work_sync() after the del_dev loop,\nand one that arrives later returns without arming. Also move INIT_WORK()\nto the top of ast_vhub_probe(), since the probe error path reuses\nast_vhub_remove() and would otherwise cancel a never-initialized work\nitem.\n\nThis issue was found by an in-house static analysis tool.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/20388c8d1d74c203fec8194c45cd31afdfab934e","tags":[]},{"url":"https://git.kernel.org/stable/c/474e9e1852fb5ba0bfb26f123dd0cb361d0f4f4d","tags":[]},{"url":"https://git.kernel.org/stable/c/c81504dfd2d93e345c98e7a012c2f2e74ce5b6a1","tags":[]},{"url":"https://git.kernel.org/stable/c/f343bfa3fc9f6255602d66f4a7dbe7ddef13ba41","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}