{"id":"CVE-2026-98864","published":"2026-10-18T16:17:40.877","lastModified":"2026-10-19T10:19:50.473","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nsched_ext: Wait for SCX_OPSS_DISPATCHING before reenqueueing a task\n\nebf1ccff79c4 (\"sched_ext: Fix ops.dequeue() semantics\") moved the final\nops_state store in scx_dispatch_enqueue() after the DSQ unlock so that the\ncustody update and ops.dequeue() precede it. A task can thus be found on a\nDSQ while still SCX_OPSS_DISPATCHING.\n\nThe dequeue and core-sched pick paths wait for the state to clear in\nops_dequeue() but the reenqueue paths don't. A reenqueue in that window runs\nops.enqueue() and sets SCX_OPSS_QUEUED before the dispatch has completed.\nThe dispatcher's final store then overwrites it with SCX_OPSS_NONE and\nfinish_dispatch() drops every later dispatch of the task.\n\nWait for SCX_OPSS_DISPATCHING to clear before dequeueing a task for\nreenqueue, the same way ops_dequeue() does.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/43457bf6550044c40bd6d3fcdf9ea66f5a55d139","tags":[]},{"url":"https://git.kernel.org/stable/c/7de9a6fb44eae4f05e68c805d58b9c618815adfa","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}